Show More
@@ -1,452 +1,459 b'' | |||||
1 | #!/usr/bin/env python |
|
1 | #!/usr/bin/env python | |
2 | # encoding: utf-8 |
|
2 | # encoding: utf-8 | |
3 | """ |
|
3 | """ | |
4 | The IPython controller application. |
|
4 | The IPython controller application. | |
5 |
|
5 | |||
6 | Authors: |
|
6 | Authors: | |
7 |
|
7 | |||
8 | * Brian Granger |
|
8 | * Brian Granger | |
9 | * MinRK |
|
9 | * MinRK | |
10 |
|
10 | |||
11 | """ |
|
11 | """ | |
12 |
|
12 | |||
13 | #----------------------------------------------------------------------------- |
|
13 | #----------------------------------------------------------------------------- | |
14 | # Copyright (C) 2008-2011 The IPython Development Team |
|
14 | # Copyright (C) 2008-2011 The IPython Development Team | |
15 | # |
|
15 | # | |
16 | # Distributed under the terms of the BSD License. The full license is in |
|
16 | # Distributed under the terms of the BSD License. The full license is in | |
17 | # the file COPYING, distributed as part of this software. |
|
17 | # the file COPYING, distributed as part of this software. | |
18 | #----------------------------------------------------------------------------- |
|
18 | #----------------------------------------------------------------------------- | |
19 |
|
19 | |||
20 | #----------------------------------------------------------------------------- |
|
20 | #----------------------------------------------------------------------------- | |
21 | # Imports |
|
21 | # Imports | |
22 | #----------------------------------------------------------------------------- |
|
22 | #----------------------------------------------------------------------------- | |
23 |
|
23 | |||
24 | from __future__ import with_statement |
|
24 | from __future__ import with_statement | |
25 |
|
25 | |||
26 | import json |
|
26 | import json | |
27 | import os |
|
27 | import os | |
28 | import socket |
|
28 | import socket | |
29 | import stat |
|
29 | import stat | |
30 | import sys |
|
30 | import sys | |
31 |
|
31 | |||
32 | from multiprocessing import Process |
|
32 | from multiprocessing import Process | |
33 |
|
33 | |||
34 | import zmq |
|
34 | import zmq | |
35 | from zmq.devices import ProcessMonitoredQueue |
|
35 | from zmq.devices import ProcessMonitoredQueue | |
36 | from zmq.log.handlers import PUBHandler |
|
36 | from zmq.log.handlers import PUBHandler | |
37 |
|
37 | |||
38 | from IPython.core.profiledir import ProfileDir |
|
38 | from IPython.core.profiledir import ProfileDir | |
39 |
|
39 | |||
40 | from IPython.parallel.apps.baseapp import ( |
|
40 | from IPython.parallel.apps.baseapp import ( | |
41 | BaseParallelApplication, |
|
41 | BaseParallelApplication, | |
42 | base_aliases, |
|
42 | base_aliases, | |
43 | base_flags, |
|
43 | base_flags, | |
44 | catch_config_error, |
|
44 | catch_config_error, | |
45 | ) |
|
45 | ) | |
46 | from IPython.utils.importstring import import_item |
|
46 | from IPython.utils.importstring import import_item | |
47 | from IPython.utils.traitlets import Instance, Unicode, Bool, List, Dict, TraitError |
|
47 | from IPython.utils.traitlets import Instance, Unicode, Bool, List, Dict, TraitError | |
48 |
|
48 | |||
49 | from IPython.zmq.session import ( |
|
49 | from IPython.zmq.session import ( | |
50 | Session, session_aliases, session_flags, default_secure |
|
50 | Session, session_aliases, session_flags, default_secure | |
51 | ) |
|
51 | ) | |
52 |
|
52 | |||
53 | from IPython.parallel.controller.heartmonitor import HeartMonitor |
|
53 | from IPython.parallel.controller.heartmonitor import HeartMonitor | |
54 | from IPython.parallel.controller.hub import HubFactory |
|
54 | from IPython.parallel.controller.hub import HubFactory | |
55 | from IPython.parallel.controller.scheduler import TaskScheduler,launch_scheduler |
|
55 | from IPython.parallel.controller.scheduler import TaskScheduler,launch_scheduler | |
56 | from IPython.parallel.controller.sqlitedb import SQLiteDB |
|
56 | from IPython.parallel.controller.sqlitedb import SQLiteDB | |
57 |
|
57 | |||
58 | from IPython.parallel.util import signal_children, split_url, disambiguate_url |
|
58 | from IPython.parallel.util import signal_children, split_url, disambiguate_url | |
59 |
|
59 | |||
60 | # conditional import of MongoDB backend class |
|
60 | # conditional import of MongoDB backend class | |
61 |
|
61 | |||
62 | try: |
|
62 | try: | |
63 | from IPython.parallel.controller.mongodb import MongoDB |
|
63 | from IPython.parallel.controller.mongodb import MongoDB | |
64 | except ImportError: |
|
64 | except ImportError: | |
65 | maybe_mongo = [] |
|
65 | maybe_mongo = [] | |
66 | else: |
|
66 | else: | |
67 | maybe_mongo = [MongoDB] |
|
67 | maybe_mongo = [MongoDB] | |
68 |
|
68 | |||
69 |
|
69 | |||
70 | #----------------------------------------------------------------------------- |
|
70 | #----------------------------------------------------------------------------- | |
71 | # Module level variables |
|
71 | # Module level variables | |
72 | #----------------------------------------------------------------------------- |
|
72 | #----------------------------------------------------------------------------- | |
73 |
|
73 | |||
74 |
|
74 | |||
75 | #: The default config file name for this application |
|
75 | #: The default config file name for this application | |
76 | default_config_file_name = u'ipcontroller_config.py' |
|
76 | default_config_file_name = u'ipcontroller_config.py' | |
77 |
|
77 | |||
78 |
|
78 | |||
79 | _description = """Start the IPython controller for parallel computing. |
|
79 | _description = """Start the IPython controller for parallel computing. | |
80 |
|
80 | |||
81 | The IPython controller provides a gateway between the IPython engines and |
|
81 | The IPython controller provides a gateway between the IPython engines and | |
82 | clients. The controller needs to be started before the engines and can be |
|
82 | clients. The controller needs to be started before the engines and can be | |
83 | configured using command line options or using a cluster directory. Cluster |
|
83 | configured using command line options or using a cluster directory. Cluster | |
84 | directories contain config, log and security files and are usually located in |
|
84 | directories contain config, log and security files and are usually located in | |
85 | your ipython directory and named as "profile_name". See the `profile` |
|
85 | your ipython directory and named as "profile_name". See the `profile` | |
86 | and `profile-dir` options for details. |
|
86 | and `profile-dir` options for details. | |
87 | """ |
|
87 | """ | |
88 |
|
88 | |||
89 | _examples = """ |
|
89 | _examples = """ | |
90 | ipcontroller --ip=192.168.0.1 --port=1000 # listen on ip, port for engines |
|
90 | ipcontroller --ip=192.168.0.1 --port=1000 # listen on ip, port for engines | |
91 | ipcontroller --scheme=pure # use the pure zeromq scheduler |
|
91 | ipcontroller --scheme=pure # use the pure zeromq scheduler | |
92 | """ |
|
92 | """ | |
93 |
|
93 | |||
94 |
|
94 | |||
95 | #----------------------------------------------------------------------------- |
|
95 | #----------------------------------------------------------------------------- | |
96 | # The main application |
|
96 | # The main application | |
97 | #----------------------------------------------------------------------------- |
|
97 | #----------------------------------------------------------------------------- | |
98 | flags = {} |
|
98 | flags = {} | |
99 | flags.update(base_flags) |
|
99 | flags.update(base_flags) | |
100 | flags.update({ |
|
100 | flags.update({ | |
101 | 'usethreads' : ( {'IPControllerApp' : {'use_threads' : True}}, |
|
101 | 'usethreads' : ( {'IPControllerApp' : {'use_threads' : True}}, | |
102 | 'Use threads instead of processes for the schedulers'), |
|
102 | 'Use threads instead of processes for the schedulers'), | |
103 | 'sqlitedb' : ({'HubFactory' : {'db_class' : 'IPython.parallel.controller.sqlitedb.SQLiteDB'}}, |
|
103 | 'sqlitedb' : ({'HubFactory' : {'db_class' : 'IPython.parallel.controller.sqlitedb.SQLiteDB'}}, | |
104 | 'use the SQLiteDB backend'), |
|
104 | 'use the SQLiteDB backend'), | |
105 | 'mongodb' : ({'HubFactory' : {'db_class' : 'IPython.parallel.controller.mongodb.MongoDB'}}, |
|
105 | 'mongodb' : ({'HubFactory' : {'db_class' : 'IPython.parallel.controller.mongodb.MongoDB'}}, | |
106 | 'use the MongoDB backend'), |
|
106 | 'use the MongoDB backend'), | |
107 | 'dictdb' : ({'HubFactory' : {'db_class' : 'IPython.parallel.controller.dictdb.DictDB'}}, |
|
107 | 'dictdb' : ({'HubFactory' : {'db_class' : 'IPython.parallel.controller.dictdb.DictDB'}}, | |
108 | 'use the in-memory DictDB backend'), |
|
108 | 'use the in-memory DictDB backend'), | |
|
109 | 'nodb' : ({'HubFactory' : {'db_class' : 'IPython.parallel.controller.dictdb.NoDB'}}, | |||
|
110 | """use dummy DB backend, which doesn't store any information. | |||
|
111 | ||||
|
112 | This can be used to prevent growth of the memory footprint of the Hub | |||
|
113 | in cases where its record-keeping is not required. Requesting results | |||
|
114 | of tasks submitted by other clients, db_queries, and task resubmission | |||
|
115 | will not be available."""), | |||
109 | 'reuse' : ({'IPControllerApp' : {'reuse_files' : True}}, |
|
116 | 'reuse' : ({'IPControllerApp' : {'reuse_files' : True}}, | |
110 | 'reuse existing json connection files') |
|
117 | 'reuse existing json connection files') | |
111 | }) |
|
118 | }) | |
112 |
|
119 | |||
113 | flags.update(session_flags) |
|
120 | flags.update(session_flags) | |
114 |
|
121 | |||
115 | aliases = dict( |
|
122 | aliases = dict( | |
116 | ssh = 'IPControllerApp.ssh_server', |
|
123 | ssh = 'IPControllerApp.ssh_server', | |
117 | enginessh = 'IPControllerApp.engine_ssh_server', |
|
124 | enginessh = 'IPControllerApp.engine_ssh_server', | |
118 | location = 'IPControllerApp.location', |
|
125 | location = 'IPControllerApp.location', | |
119 |
|
126 | |||
120 | url = 'HubFactory.url', |
|
127 | url = 'HubFactory.url', | |
121 | ip = 'HubFactory.ip', |
|
128 | ip = 'HubFactory.ip', | |
122 | transport = 'HubFactory.transport', |
|
129 | transport = 'HubFactory.transport', | |
123 | port = 'HubFactory.regport', |
|
130 | port = 'HubFactory.regport', | |
124 |
|
131 | |||
125 | ping = 'HeartMonitor.period', |
|
132 | ping = 'HeartMonitor.period', | |
126 |
|
133 | |||
127 | scheme = 'TaskScheduler.scheme_name', |
|
134 | scheme = 'TaskScheduler.scheme_name', | |
128 | hwm = 'TaskScheduler.hwm', |
|
135 | hwm = 'TaskScheduler.hwm', | |
129 | ) |
|
136 | ) | |
130 | aliases.update(base_aliases) |
|
137 | aliases.update(base_aliases) | |
131 | aliases.update(session_aliases) |
|
138 | aliases.update(session_aliases) | |
132 |
|
139 | |||
133 |
|
140 | |||
134 | class IPControllerApp(BaseParallelApplication): |
|
141 | class IPControllerApp(BaseParallelApplication): | |
135 |
|
142 | |||
136 | name = u'ipcontroller' |
|
143 | name = u'ipcontroller' | |
137 | description = _description |
|
144 | description = _description | |
138 | examples = _examples |
|
145 | examples = _examples | |
139 | config_file_name = Unicode(default_config_file_name) |
|
146 | config_file_name = Unicode(default_config_file_name) | |
140 | classes = [ProfileDir, Session, HubFactory, TaskScheduler, HeartMonitor, SQLiteDB] + maybe_mongo |
|
147 | classes = [ProfileDir, Session, HubFactory, TaskScheduler, HeartMonitor, SQLiteDB] + maybe_mongo | |
141 |
|
148 | |||
142 | # change default to True |
|
149 | # change default to True | |
143 | auto_create = Bool(True, config=True, |
|
150 | auto_create = Bool(True, config=True, | |
144 | help="""Whether to create profile dir if it doesn't exist.""") |
|
151 | help="""Whether to create profile dir if it doesn't exist.""") | |
145 |
|
152 | |||
146 | reuse_files = Bool(False, config=True, |
|
153 | reuse_files = Bool(False, config=True, | |
147 | help='Whether to reuse existing json connection files.' |
|
154 | help='Whether to reuse existing json connection files.' | |
148 | ) |
|
155 | ) | |
149 | ssh_server = Unicode(u'', config=True, |
|
156 | ssh_server = Unicode(u'', config=True, | |
150 | help="""ssh url for clients to use when connecting to the Controller |
|
157 | help="""ssh url for clients to use when connecting to the Controller | |
151 | processes. It should be of the form: [user@]server[:port]. The |
|
158 | processes. It should be of the form: [user@]server[:port]. The | |
152 | Controller's listening addresses must be accessible from the ssh server""", |
|
159 | Controller's listening addresses must be accessible from the ssh server""", | |
153 | ) |
|
160 | ) | |
154 | engine_ssh_server = Unicode(u'', config=True, |
|
161 | engine_ssh_server = Unicode(u'', config=True, | |
155 | help="""ssh url for engines to use when connecting to the Controller |
|
162 | help="""ssh url for engines to use when connecting to the Controller | |
156 | processes. It should be of the form: [user@]server[:port]. The |
|
163 | processes. It should be of the form: [user@]server[:port]. The | |
157 | Controller's listening addresses must be accessible from the ssh server""", |
|
164 | Controller's listening addresses must be accessible from the ssh server""", | |
158 | ) |
|
165 | ) | |
159 | location = Unicode(u'', config=True, |
|
166 | location = Unicode(u'', config=True, | |
160 | help="""The external IP or domain name of the Controller, used for disambiguating |
|
167 | help="""The external IP or domain name of the Controller, used for disambiguating | |
161 | engine and client connections.""", |
|
168 | engine and client connections.""", | |
162 | ) |
|
169 | ) | |
163 | import_statements = List([], config=True, |
|
170 | import_statements = List([], config=True, | |
164 | help="import statements to be run at startup. Necessary in some environments" |
|
171 | help="import statements to be run at startup. Necessary in some environments" | |
165 | ) |
|
172 | ) | |
166 |
|
173 | |||
167 | use_threads = Bool(False, config=True, |
|
174 | use_threads = Bool(False, config=True, | |
168 | help='Use threads instead of processes for the schedulers', |
|
175 | help='Use threads instead of processes for the schedulers', | |
169 | ) |
|
176 | ) | |
170 |
|
177 | |||
171 | engine_json_file = Unicode('ipcontroller-engine.json', config=True, |
|
178 | engine_json_file = Unicode('ipcontroller-engine.json', config=True, | |
172 | help="JSON filename where engine connection info will be stored.") |
|
179 | help="JSON filename where engine connection info will be stored.") | |
173 | client_json_file = Unicode('ipcontroller-client.json', config=True, |
|
180 | client_json_file = Unicode('ipcontroller-client.json', config=True, | |
174 | help="JSON filename where client connection info will be stored.") |
|
181 | help="JSON filename where client connection info will be stored.") | |
175 |
|
182 | |||
176 | def _cluster_id_changed(self, name, old, new): |
|
183 | def _cluster_id_changed(self, name, old, new): | |
177 | super(IPControllerApp, self)._cluster_id_changed(name, old, new) |
|
184 | super(IPControllerApp, self)._cluster_id_changed(name, old, new) | |
178 | self.engine_json_file = "%s-engine.json" % self.name |
|
185 | self.engine_json_file = "%s-engine.json" % self.name | |
179 | self.client_json_file = "%s-client.json" % self.name |
|
186 | self.client_json_file = "%s-client.json" % self.name | |
180 |
|
187 | |||
181 |
|
188 | |||
182 | # internal |
|
189 | # internal | |
183 | children = List() |
|
190 | children = List() | |
184 | mq_class = Unicode('zmq.devices.ProcessMonitoredQueue') |
|
191 | mq_class = Unicode('zmq.devices.ProcessMonitoredQueue') | |
185 |
|
192 | |||
186 | def _use_threads_changed(self, name, old, new): |
|
193 | def _use_threads_changed(self, name, old, new): | |
187 | self.mq_class = 'zmq.devices.%sMonitoredQueue'%('Thread' if new else 'Process') |
|
194 | self.mq_class = 'zmq.devices.%sMonitoredQueue'%('Thread' if new else 'Process') | |
188 |
|
195 | |||
189 | aliases = Dict(aliases) |
|
196 | aliases = Dict(aliases) | |
190 | flags = Dict(flags) |
|
197 | flags = Dict(flags) | |
191 |
|
198 | |||
192 |
|
199 | |||
193 | def save_connection_dict(self, fname, cdict): |
|
200 | def save_connection_dict(self, fname, cdict): | |
194 | """save a connection dict to json file.""" |
|
201 | """save a connection dict to json file.""" | |
195 | c = self.config |
|
202 | c = self.config | |
196 | url = cdict['url'] |
|
203 | url = cdict['url'] | |
197 | location = cdict['location'] |
|
204 | location = cdict['location'] | |
198 | if not location: |
|
205 | if not location: | |
199 | try: |
|
206 | try: | |
200 | proto,ip,port = split_url(url) |
|
207 | proto,ip,port = split_url(url) | |
201 | except AssertionError: |
|
208 | except AssertionError: | |
202 | pass |
|
209 | pass | |
203 | else: |
|
210 | else: | |
204 | try: |
|
211 | try: | |
205 | location = socket.gethostbyname_ex(socket.gethostname())[2][-1] |
|
212 | location = socket.gethostbyname_ex(socket.gethostname())[2][-1] | |
206 | except (socket.gaierror, IndexError): |
|
213 | except (socket.gaierror, IndexError): | |
207 | self.log.warn("Could not identify this machine's IP, assuming 127.0.0.1." |
|
214 | self.log.warn("Could not identify this machine's IP, assuming 127.0.0.1." | |
208 | " You may need to specify '--location=<external_ip_address>' to help" |
|
215 | " You may need to specify '--location=<external_ip_address>' to help" | |
209 | " IPython decide when to connect via loopback.") |
|
216 | " IPython decide when to connect via loopback.") | |
210 | location = '127.0.0.1' |
|
217 | location = '127.0.0.1' | |
211 | cdict['location'] = location |
|
218 | cdict['location'] = location | |
212 | fname = os.path.join(self.profile_dir.security_dir, fname) |
|
219 | fname = os.path.join(self.profile_dir.security_dir, fname) | |
213 | self.log.info("writing connection info to %s", fname) |
|
220 | self.log.info("writing connection info to %s", fname) | |
214 | with open(fname, 'w') as f: |
|
221 | with open(fname, 'w') as f: | |
215 | f.write(json.dumps(cdict, indent=2)) |
|
222 | f.write(json.dumps(cdict, indent=2)) | |
216 | os.chmod(fname, stat.S_IRUSR|stat.S_IWUSR) |
|
223 | os.chmod(fname, stat.S_IRUSR|stat.S_IWUSR) | |
217 |
|
224 | |||
218 | def load_config_from_json(self): |
|
225 | def load_config_from_json(self): | |
219 | """load config from existing json connector files.""" |
|
226 | """load config from existing json connector files.""" | |
220 | c = self.config |
|
227 | c = self.config | |
221 | self.log.debug("loading config from JSON") |
|
228 | self.log.debug("loading config from JSON") | |
222 | # load from engine config |
|
229 | # load from engine config | |
223 | fname = os.path.join(self.profile_dir.security_dir, self.engine_json_file) |
|
230 | fname = os.path.join(self.profile_dir.security_dir, self.engine_json_file) | |
224 | self.log.info("loading connection info from %s", fname) |
|
231 | self.log.info("loading connection info from %s", fname) | |
225 | with open(fname) as f: |
|
232 | with open(fname) as f: | |
226 | cfg = json.loads(f.read()) |
|
233 | cfg = json.loads(f.read()) | |
227 | key = cfg['exec_key'] |
|
234 | key = cfg['exec_key'] | |
228 | # json gives unicode, Session.key wants bytes |
|
235 | # json gives unicode, Session.key wants bytes | |
229 | c.Session.key = key.encode('ascii') |
|
236 | c.Session.key = key.encode('ascii') | |
230 | xport,addr = cfg['url'].split('://') |
|
237 | xport,addr = cfg['url'].split('://') | |
231 | c.HubFactory.engine_transport = xport |
|
238 | c.HubFactory.engine_transport = xport | |
232 | ip,ports = addr.split(':') |
|
239 | ip,ports = addr.split(':') | |
233 | c.HubFactory.engine_ip = ip |
|
240 | c.HubFactory.engine_ip = ip | |
234 | c.HubFactory.regport = int(ports) |
|
241 | c.HubFactory.regport = int(ports) | |
235 | self.location = cfg['location'] |
|
242 | self.location = cfg['location'] | |
236 | if not self.engine_ssh_server: |
|
243 | if not self.engine_ssh_server: | |
237 | self.engine_ssh_server = cfg['ssh'] |
|
244 | self.engine_ssh_server = cfg['ssh'] | |
238 | # load client config |
|
245 | # load client config | |
239 | fname = os.path.join(self.profile_dir.security_dir, self.client_json_file) |
|
246 | fname = os.path.join(self.profile_dir.security_dir, self.client_json_file) | |
240 | self.log.info("loading connection info from %s", fname) |
|
247 | self.log.info("loading connection info from %s", fname) | |
241 | with open(fname) as f: |
|
248 | with open(fname) as f: | |
242 | cfg = json.loads(f.read()) |
|
249 | cfg = json.loads(f.read()) | |
243 | assert key == cfg['exec_key'], "exec_key mismatch between engine and client keys" |
|
250 | assert key == cfg['exec_key'], "exec_key mismatch between engine and client keys" | |
244 | xport,addr = cfg['url'].split('://') |
|
251 | xport,addr = cfg['url'].split('://') | |
245 | c.HubFactory.client_transport = xport |
|
252 | c.HubFactory.client_transport = xport | |
246 | ip,ports = addr.split(':') |
|
253 | ip,ports = addr.split(':') | |
247 | c.HubFactory.client_ip = ip |
|
254 | c.HubFactory.client_ip = ip | |
248 | if not self.ssh_server: |
|
255 | if not self.ssh_server: | |
249 | self.ssh_server = cfg['ssh'] |
|
256 | self.ssh_server = cfg['ssh'] | |
250 | assert int(ports) == c.HubFactory.regport, "regport mismatch" |
|
257 | assert int(ports) == c.HubFactory.regport, "regport mismatch" | |
251 |
|
258 | |||
252 | def load_secondary_config(self): |
|
259 | def load_secondary_config(self): | |
253 | """secondary config, loading from JSON and setting defaults""" |
|
260 | """secondary config, loading from JSON and setting defaults""" | |
254 | if self.reuse_files: |
|
261 | if self.reuse_files: | |
255 | try: |
|
262 | try: | |
256 | self.load_config_from_json() |
|
263 | self.load_config_from_json() | |
257 | except (AssertionError,IOError) as e: |
|
264 | except (AssertionError,IOError) as e: | |
258 | self.log.error("Could not load config from JSON: %s" % e) |
|
265 | self.log.error("Could not load config from JSON: %s" % e) | |
259 | self.reuse_files=False |
|
266 | self.reuse_files=False | |
260 | # switch Session.key default to secure |
|
267 | # switch Session.key default to secure | |
261 | default_secure(self.config) |
|
268 | default_secure(self.config) | |
262 | self.log.debug("Config changed") |
|
269 | self.log.debug("Config changed") | |
263 | self.log.debug(repr(self.config)) |
|
270 | self.log.debug(repr(self.config)) | |
264 |
|
271 | |||
265 | def init_hub(self): |
|
272 | def init_hub(self): | |
266 | c = self.config |
|
273 | c = self.config | |
267 |
|
274 | |||
268 | self.do_import_statements() |
|
275 | self.do_import_statements() | |
269 |
|
276 | |||
270 | try: |
|
277 | try: | |
271 | self.factory = HubFactory(config=c, log=self.log) |
|
278 | self.factory = HubFactory(config=c, log=self.log) | |
272 | # self.start_logging() |
|
279 | # self.start_logging() | |
273 | self.factory.init_hub() |
|
280 | self.factory.init_hub() | |
274 | except TraitError: |
|
281 | except TraitError: | |
275 | raise |
|
282 | raise | |
276 | except Exception: |
|
283 | except Exception: | |
277 | self.log.error("Couldn't construct the Controller", exc_info=True) |
|
284 | self.log.error("Couldn't construct the Controller", exc_info=True) | |
278 | self.exit(1) |
|
285 | self.exit(1) | |
279 |
|
286 | |||
280 | if not self.reuse_files: |
|
287 | if not self.reuse_files: | |
281 | # save to new json config files |
|
288 | # save to new json config files | |
282 | f = self.factory |
|
289 | f = self.factory | |
283 | cdict = {'exec_key' : f.session.key.decode('ascii'), |
|
290 | cdict = {'exec_key' : f.session.key.decode('ascii'), | |
284 | 'ssh' : self.ssh_server, |
|
291 | 'ssh' : self.ssh_server, | |
285 | 'url' : "%s://%s:%s"%(f.client_transport, f.client_ip, f.regport), |
|
292 | 'url' : "%s://%s:%s"%(f.client_transport, f.client_ip, f.regport), | |
286 | 'location' : self.location |
|
293 | 'location' : self.location | |
287 | } |
|
294 | } | |
288 | self.save_connection_dict(self.client_json_file, cdict) |
|
295 | self.save_connection_dict(self.client_json_file, cdict) | |
289 | edict = cdict |
|
296 | edict = cdict | |
290 | edict['url']="%s://%s:%s"%((f.client_transport, f.client_ip, f.regport)) |
|
297 | edict['url']="%s://%s:%s"%((f.client_transport, f.client_ip, f.regport)) | |
291 | edict['ssh'] = self.engine_ssh_server |
|
298 | edict['ssh'] = self.engine_ssh_server | |
292 | self.save_connection_dict(self.engine_json_file, edict) |
|
299 | self.save_connection_dict(self.engine_json_file, edict) | |
293 |
|
300 | |||
294 | # |
|
301 | # | |
295 | def init_schedulers(self): |
|
302 | def init_schedulers(self): | |
296 | children = self.children |
|
303 | children = self.children | |
297 | mq = import_item(str(self.mq_class)) |
|
304 | mq = import_item(str(self.mq_class)) | |
298 |
|
305 | |||
299 | hub = self.factory |
|
306 | hub = self.factory | |
300 | # disambiguate url, in case of * |
|
307 | # disambiguate url, in case of * | |
301 | monitor_url = disambiguate_url(hub.monitor_url) |
|
308 | monitor_url = disambiguate_url(hub.monitor_url) | |
302 | # maybe_inproc = 'inproc://monitor' if self.use_threads else monitor_url |
|
309 | # maybe_inproc = 'inproc://monitor' if self.use_threads else monitor_url | |
303 | # IOPub relay (in a Process) |
|
310 | # IOPub relay (in a Process) | |
304 | q = mq(zmq.PUB, zmq.SUB, zmq.PUB, b'N/A',b'iopub') |
|
311 | q = mq(zmq.PUB, zmq.SUB, zmq.PUB, b'N/A',b'iopub') | |
305 | q.bind_in(hub.client_info['iopub']) |
|
312 | q.bind_in(hub.client_info['iopub']) | |
306 | q.bind_out(hub.engine_info['iopub']) |
|
313 | q.bind_out(hub.engine_info['iopub']) | |
307 | q.setsockopt_out(zmq.SUBSCRIBE, b'') |
|
314 | q.setsockopt_out(zmq.SUBSCRIBE, b'') | |
308 | q.connect_mon(monitor_url) |
|
315 | q.connect_mon(monitor_url) | |
309 | q.daemon=True |
|
316 | q.daemon=True | |
310 | children.append(q) |
|
317 | children.append(q) | |
311 |
|
318 | |||
312 | # Multiplexer Queue (in a Process) |
|
319 | # Multiplexer Queue (in a Process) | |
313 | q = mq(zmq.ROUTER, zmq.ROUTER, zmq.PUB, b'in', b'out') |
|
320 | q = mq(zmq.ROUTER, zmq.ROUTER, zmq.PUB, b'in', b'out') | |
314 | q.bind_in(hub.client_info['mux']) |
|
321 | q.bind_in(hub.client_info['mux']) | |
315 | q.setsockopt_in(zmq.IDENTITY, b'mux') |
|
322 | q.setsockopt_in(zmq.IDENTITY, b'mux') | |
316 | q.bind_out(hub.engine_info['mux']) |
|
323 | q.bind_out(hub.engine_info['mux']) | |
317 | q.connect_mon(monitor_url) |
|
324 | q.connect_mon(monitor_url) | |
318 | q.daemon=True |
|
325 | q.daemon=True | |
319 | children.append(q) |
|
326 | children.append(q) | |
320 |
|
327 | |||
321 | # Control Queue (in a Process) |
|
328 | # Control Queue (in a Process) | |
322 | q = mq(zmq.ROUTER, zmq.ROUTER, zmq.PUB, b'incontrol', b'outcontrol') |
|
329 | q = mq(zmq.ROUTER, zmq.ROUTER, zmq.PUB, b'incontrol', b'outcontrol') | |
323 | q.bind_in(hub.client_info['control']) |
|
330 | q.bind_in(hub.client_info['control']) | |
324 | q.setsockopt_in(zmq.IDENTITY, b'control') |
|
331 | q.setsockopt_in(zmq.IDENTITY, b'control') | |
325 | q.bind_out(hub.engine_info['control']) |
|
332 | q.bind_out(hub.engine_info['control']) | |
326 | q.connect_mon(monitor_url) |
|
333 | q.connect_mon(monitor_url) | |
327 | q.daemon=True |
|
334 | q.daemon=True | |
328 | children.append(q) |
|
335 | children.append(q) | |
329 | try: |
|
336 | try: | |
330 | scheme = self.config.TaskScheduler.scheme_name |
|
337 | scheme = self.config.TaskScheduler.scheme_name | |
331 | except AttributeError: |
|
338 | except AttributeError: | |
332 | scheme = TaskScheduler.scheme_name.get_default_value() |
|
339 | scheme = TaskScheduler.scheme_name.get_default_value() | |
333 | # Task Queue (in a Process) |
|
340 | # Task Queue (in a Process) | |
334 | if scheme == 'pure': |
|
341 | if scheme == 'pure': | |
335 | self.log.warn("task::using pure XREQ Task scheduler") |
|
342 | self.log.warn("task::using pure XREQ Task scheduler") | |
336 | q = mq(zmq.ROUTER, zmq.DEALER, zmq.PUB, b'intask', b'outtask') |
|
343 | q = mq(zmq.ROUTER, zmq.DEALER, zmq.PUB, b'intask', b'outtask') | |
337 | # q.setsockopt_out(zmq.HWM, hub.hwm) |
|
344 | # q.setsockopt_out(zmq.HWM, hub.hwm) | |
338 | q.bind_in(hub.client_info['task'][1]) |
|
345 | q.bind_in(hub.client_info['task'][1]) | |
339 | q.setsockopt_in(zmq.IDENTITY, b'task') |
|
346 | q.setsockopt_in(zmq.IDENTITY, b'task') | |
340 | q.bind_out(hub.engine_info['task']) |
|
347 | q.bind_out(hub.engine_info['task']) | |
341 | q.connect_mon(monitor_url) |
|
348 | q.connect_mon(monitor_url) | |
342 | q.daemon=True |
|
349 | q.daemon=True | |
343 | children.append(q) |
|
350 | children.append(q) | |
344 | elif scheme == 'none': |
|
351 | elif scheme == 'none': | |
345 | self.log.warn("task::using no Task scheduler") |
|
352 | self.log.warn("task::using no Task scheduler") | |
346 |
|
353 | |||
347 | else: |
|
354 | else: | |
348 | self.log.info("task::using Python %s Task scheduler"%scheme) |
|
355 | self.log.info("task::using Python %s Task scheduler"%scheme) | |
349 | sargs = (hub.client_info['task'][1], hub.engine_info['task'], |
|
356 | sargs = (hub.client_info['task'][1], hub.engine_info['task'], | |
350 | monitor_url, disambiguate_url(hub.client_info['notification'])) |
|
357 | monitor_url, disambiguate_url(hub.client_info['notification'])) | |
351 | kwargs = dict(logname='scheduler', loglevel=self.log_level, |
|
358 | kwargs = dict(logname='scheduler', loglevel=self.log_level, | |
352 | log_url = self.log_url, config=dict(self.config)) |
|
359 | log_url = self.log_url, config=dict(self.config)) | |
353 | if 'Process' in self.mq_class: |
|
360 | if 'Process' in self.mq_class: | |
354 | # run the Python scheduler in a Process |
|
361 | # run the Python scheduler in a Process | |
355 | q = Process(target=launch_scheduler, args=sargs, kwargs=kwargs) |
|
362 | q = Process(target=launch_scheduler, args=sargs, kwargs=kwargs) | |
356 | q.daemon=True |
|
363 | q.daemon=True | |
357 | children.append(q) |
|
364 | children.append(q) | |
358 | else: |
|
365 | else: | |
359 | # single-threaded Controller |
|
366 | # single-threaded Controller | |
360 | kwargs['in_thread'] = True |
|
367 | kwargs['in_thread'] = True | |
361 | launch_scheduler(*sargs, **kwargs) |
|
368 | launch_scheduler(*sargs, **kwargs) | |
362 |
|
369 | |||
363 |
|
370 | |||
364 | def save_urls(self): |
|
371 | def save_urls(self): | |
365 | """save the registration urls to files.""" |
|
372 | """save the registration urls to files.""" | |
366 | c = self.config |
|
373 | c = self.config | |
367 |
|
374 | |||
368 | sec_dir = self.profile_dir.security_dir |
|
375 | sec_dir = self.profile_dir.security_dir | |
369 | cf = self.factory |
|
376 | cf = self.factory | |
370 |
|
377 | |||
371 | with open(os.path.join(sec_dir, 'ipcontroller-engine.url'), 'w') as f: |
|
378 | with open(os.path.join(sec_dir, 'ipcontroller-engine.url'), 'w') as f: | |
372 | f.write("%s://%s:%s"%(cf.engine_transport, cf.engine_ip, cf.regport)) |
|
379 | f.write("%s://%s:%s"%(cf.engine_transport, cf.engine_ip, cf.regport)) | |
373 |
|
380 | |||
374 | with open(os.path.join(sec_dir, 'ipcontroller-client.url'), 'w') as f: |
|
381 | with open(os.path.join(sec_dir, 'ipcontroller-client.url'), 'w') as f: | |
375 | f.write("%s://%s:%s"%(cf.client_transport, cf.client_ip, cf.regport)) |
|
382 | f.write("%s://%s:%s"%(cf.client_transport, cf.client_ip, cf.regport)) | |
376 |
|
383 | |||
377 |
|
384 | |||
378 | def do_import_statements(self): |
|
385 | def do_import_statements(self): | |
379 | statements = self.import_statements |
|
386 | statements = self.import_statements | |
380 | for s in statements: |
|
387 | for s in statements: | |
381 | try: |
|
388 | try: | |
382 | self.log.msg("Executing statement: '%s'" % s) |
|
389 | self.log.msg("Executing statement: '%s'" % s) | |
383 | exec s in globals(), locals() |
|
390 | exec s in globals(), locals() | |
384 | except: |
|
391 | except: | |
385 | self.log.msg("Error running statement: %s" % s) |
|
392 | self.log.msg("Error running statement: %s" % s) | |
386 |
|
393 | |||
387 | def forward_logging(self): |
|
394 | def forward_logging(self): | |
388 | if self.log_url: |
|
395 | if self.log_url: | |
389 | self.log.info("Forwarding logging to %s"%self.log_url) |
|
396 | self.log.info("Forwarding logging to %s"%self.log_url) | |
390 | context = zmq.Context.instance() |
|
397 | context = zmq.Context.instance() | |
391 | lsock = context.socket(zmq.PUB) |
|
398 | lsock = context.socket(zmq.PUB) | |
392 | lsock.connect(self.log_url) |
|
399 | lsock.connect(self.log_url) | |
393 | handler = PUBHandler(lsock) |
|
400 | handler = PUBHandler(lsock) | |
394 | self.log.removeHandler(self._log_handler) |
|
401 | self.log.removeHandler(self._log_handler) | |
395 | handler.root_topic = 'controller' |
|
402 | handler.root_topic = 'controller' | |
396 | handler.setLevel(self.log_level) |
|
403 | handler.setLevel(self.log_level) | |
397 | self.log.addHandler(handler) |
|
404 | self.log.addHandler(handler) | |
398 | self._log_handler = handler |
|
405 | self._log_handler = handler | |
399 |
|
406 | |||
400 | @catch_config_error |
|
407 | @catch_config_error | |
401 | def initialize(self, argv=None): |
|
408 | def initialize(self, argv=None): | |
402 | super(IPControllerApp, self).initialize(argv) |
|
409 | super(IPControllerApp, self).initialize(argv) | |
403 | self.forward_logging() |
|
410 | self.forward_logging() | |
404 | self.load_secondary_config() |
|
411 | self.load_secondary_config() | |
405 | self.init_hub() |
|
412 | self.init_hub() | |
406 | self.init_schedulers() |
|
413 | self.init_schedulers() | |
407 |
|
414 | |||
408 | def start(self): |
|
415 | def start(self): | |
409 | # Start the subprocesses: |
|
416 | # Start the subprocesses: | |
410 | self.factory.start() |
|
417 | self.factory.start() | |
411 | child_procs = [] |
|
418 | child_procs = [] | |
412 | for child in self.children: |
|
419 | for child in self.children: | |
413 | child.start() |
|
420 | child.start() | |
414 | if isinstance(child, ProcessMonitoredQueue): |
|
421 | if isinstance(child, ProcessMonitoredQueue): | |
415 | child_procs.append(child.launcher) |
|
422 | child_procs.append(child.launcher) | |
416 | elif isinstance(child, Process): |
|
423 | elif isinstance(child, Process): | |
417 | child_procs.append(child) |
|
424 | child_procs.append(child) | |
418 | if child_procs: |
|
425 | if child_procs: | |
419 | signal_children(child_procs) |
|
426 | signal_children(child_procs) | |
420 |
|
427 | |||
421 | self.write_pid_file(overwrite=True) |
|
428 | self.write_pid_file(overwrite=True) | |
422 |
|
429 | |||
423 | try: |
|
430 | try: | |
424 | self.factory.loop.start() |
|
431 | self.factory.loop.start() | |
425 | except KeyboardInterrupt: |
|
432 | except KeyboardInterrupt: | |
426 | self.log.critical("Interrupted, Exiting...\n") |
|
433 | self.log.critical("Interrupted, Exiting...\n") | |
427 |
|
434 | |||
428 |
|
435 | |||
429 |
|
436 | |||
430 | def launch_new_instance(): |
|
437 | def launch_new_instance(): | |
431 | """Create and run the IPython controller""" |
|
438 | """Create and run the IPython controller""" | |
432 | if sys.platform == 'win32': |
|
439 | if sys.platform == 'win32': | |
433 | # make sure we don't get called from a multiprocessing subprocess |
|
440 | # make sure we don't get called from a multiprocessing subprocess | |
434 | # this can result in infinite Controllers being started on Windows |
|
441 | # this can result in infinite Controllers being started on Windows | |
435 | # which doesn't have a proper fork, so multiprocessing is wonky |
|
442 | # which doesn't have a proper fork, so multiprocessing is wonky | |
436 |
|
443 | |||
437 | # this only comes up when IPython has been installed using vanilla |
|
444 | # this only comes up when IPython has been installed using vanilla | |
438 | # setuptools, and *not* distribute. |
|
445 | # setuptools, and *not* distribute. | |
439 | import multiprocessing |
|
446 | import multiprocessing | |
440 | p = multiprocessing.current_process() |
|
447 | p = multiprocessing.current_process() | |
441 | # the main process has name 'MainProcess' |
|
448 | # the main process has name 'MainProcess' | |
442 | # subprocesses will have names like 'Process-1' |
|
449 | # subprocesses will have names like 'Process-1' | |
443 | if p.name != 'MainProcess': |
|
450 | if p.name != 'MainProcess': | |
444 | # we are a subprocess, don't start another Controller! |
|
451 | # we are a subprocess, don't start another Controller! | |
445 | return |
|
452 | return | |
446 | app = IPControllerApp.instance() |
|
453 | app = IPControllerApp.instance() | |
447 | app.initialize() |
|
454 | app.initialize() | |
448 | app.start() |
|
455 | app.start() | |
449 |
|
456 | |||
450 |
|
457 | |||
451 | if __name__ == '__main__': |
|
458 | if __name__ == '__main__': | |
452 | launch_new_instance() |
|
459 | launch_new_instance() |
@@ -1,185 +1,216 b'' | |||||
1 | """A Task logger that presents our DB interface, |
|
1 | """A Task logger that presents our DB interface, | |
2 | but exists entirely in memory and implemented with dicts. |
|
2 | but exists entirely in memory and implemented with dicts. | |
3 |
|
3 | |||
4 | Authors: |
|
4 | Authors: | |
5 |
|
5 | |||
6 | * Min RK |
|
6 | * Min RK | |
7 |
|
7 | |||
8 |
|
8 | |||
9 | TaskRecords are dicts of the form: |
|
9 | TaskRecords are dicts of the form: | |
10 | { |
|
10 | { | |
11 | 'msg_id' : str(uuid), |
|
11 | 'msg_id' : str(uuid), | |
12 | 'client_uuid' : str(uuid), |
|
12 | 'client_uuid' : str(uuid), | |
13 | 'engine_uuid' : str(uuid) or None, |
|
13 | 'engine_uuid' : str(uuid) or None, | |
14 | 'header' : dict(header), |
|
14 | 'header' : dict(header), | |
15 | 'content': dict(content), |
|
15 | 'content': dict(content), | |
16 | 'buffers': list(buffers), |
|
16 | 'buffers': list(buffers), | |
17 | 'submitted': datetime, |
|
17 | 'submitted': datetime, | |
18 | 'started': datetime or None, |
|
18 | 'started': datetime or None, | |
19 | 'completed': datetime or None, |
|
19 | 'completed': datetime or None, | |
20 | 'resubmitted': datetime or None, |
|
20 | 'resubmitted': datetime or None, | |
21 | 'result_header' : dict(header) or None, |
|
21 | 'result_header' : dict(header) or None, | |
22 | 'result_content' : dict(content) or None, |
|
22 | 'result_content' : dict(content) or None, | |
23 | 'result_buffers' : list(buffers) or None, |
|
23 | 'result_buffers' : list(buffers) or None, | |
24 | } |
|
24 | } | |
25 | With this info, many of the special categories of tasks can be defined by query: |
|
25 | With this info, many of the special categories of tasks can be defined by query: | |
26 |
|
26 | |||
27 | pending: completed is None |
|
27 | pending: completed is None | |
28 | client's outstanding: client_uuid = uuid && completed is None |
|
28 | client's outstanding: client_uuid = uuid && completed is None | |
29 | MIA: arrived is None (and completed is None) |
|
29 | MIA: arrived is None (and completed is None) | |
30 | etc. |
|
30 | etc. | |
31 |
|
31 | |||
32 | EngineRecords are dicts of the form: |
|
32 | EngineRecords are dicts of the form: | |
33 | { |
|
33 | { | |
34 | 'eid' : int(id), |
|
34 | 'eid' : int(id), | |
35 | 'uuid': str(uuid) |
|
35 | 'uuid': str(uuid) | |
36 | } |
|
36 | } | |
37 | This may be extended, but is currently. |
|
37 | This may be extended, but is currently. | |
38 |
|
38 | |||
39 | We support a subset of mongodb operators: |
|
39 | We support a subset of mongodb operators: | |
40 | $lt,$gt,$lte,$gte,$ne,$in,$nin,$all,$mod,$exists |
|
40 | $lt,$gt,$lte,$gte,$ne,$in,$nin,$all,$mod,$exists | |
41 | """ |
|
41 | """ | |
42 | #----------------------------------------------------------------------------- |
|
42 | #----------------------------------------------------------------------------- | |
43 | # Copyright (C) 2010-2011 The IPython Development Team |
|
43 | # Copyright (C) 2010-2011 The IPython Development Team | |
44 | # |
|
44 | # | |
45 | # Distributed under the terms of the BSD License. The full license is in |
|
45 | # Distributed under the terms of the BSD License. The full license is in | |
46 | # the file COPYING, distributed as part of this software. |
|
46 | # the file COPYING, distributed as part of this software. | |
47 | #----------------------------------------------------------------------------- |
|
47 | #----------------------------------------------------------------------------- | |
48 |
|
48 | |||
49 |
|
49 | |||
50 | from datetime import datetime |
|
50 | from datetime import datetime | |
51 |
|
51 | |||
52 | from IPython.config.configurable import LoggingConfigurable |
|
52 | from IPython.config.configurable import LoggingConfigurable | |
53 |
|
53 | |||
54 | from IPython.utils.traitlets import Dict, Unicode, Instance |
|
54 | from IPython.utils.traitlets import Dict, Unicode, Instance | |
55 |
|
55 | |||
56 | filters = { |
|
56 | filters = { | |
57 | '$lt' : lambda a,b: a < b, |
|
57 | '$lt' : lambda a,b: a < b, | |
58 | '$gt' : lambda a,b: b > a, |
|
58 | '$gt' : lambda a,b: b > a, | |
59 | '$eq' : lambda a,b: a == b, |
|
59 | '$eq' : lambda a,b: a == b, | |
60 | '$ne' : lambda a,b: a != b, |
|
60 | '$ne' : lambda a,b: a != b, | |
61 | '$lte': lambda a,b: a <= b, |
|
61 | '$lte': lambda a,b: a <= b, | |
62 | '$gte': lambda a,b: a >= b, |
|
62 | '$gte': lambda a,b: a >= b, | |
63 | '$in' : lambda a,b: a in b, |
|
63 | '$in' : lambda a,b: a in b, | |
64 | '$nin': lambda a,b: a not in b, |
|
64 | '$nin': lambda a,b: a not in b, | |
65 | '$all': lambda a,b: all([ a in bb for bb in b ]), |
|
65 | '$all': lambda a,b: all([ a in bb for bb in b ]), | |
66 | '$mod': lambda a,b: a%b[0] == b[1], |
|
66 | '$mod': lambda a,b: a%b[0] == b[1], | |
67 | '$exists' : lambda a,b: (b and a is not None) or (a is None and not b) |
|
67 | '$exists' : lambda a,b: (b and a is not None) or (a is None and not b) | |
68 | } |
|
68 | } | |
69 |
|
69 | |||
70 |
|
70 | |||
71 | class CompositeFilter(object): |
|
71 | class CompositeFilter(object): | |
72 | """Composite filter for matching multiple properties.""" |
|
72 | """Composite filter for matching multiple properties.""" | |
73 |
|
73 | |||
74 | def __init__(self, dikt): |
|
74 | def __init__(self, dikt): | |
75 | self.tests = [] |
|
75 | self.tests = [] | |
76 | self.values = [] |
|
76 | self.values = [] | |
77 | for key, value in dikt.iteritems(): |
|
77 | for key, value in dikt.iteritems(): | |
78 | self.tests.append(filters[key]) |
|
78 | self.tests.append(filters[key]) | |
79 | self.values.append(value) |
|
79 | self.values.append(value) | |
80 |
|
80 | |||
81 | def __call__(self, value): |
|
81 | def __call__(self, value): | |
82 | for test,check in zip(self.tests, self.values): |
|
82 | for test,check in zip(self.tests, self.values): | |
83 | if not test(value, check): |
|
83 | if not test(value, check): | |
84 | return False |
|
84 | return False | |
85 | return True |
|
85 | return True | |
86 |
|
86 | |||
87 | class BaseDB(LoggingConfigurable): |
|
87 | class BaseDB(LoggingConfigurable): | |
88 | """Empty Parent class so traitlets work on DB.""" |
|
88 | """Empty Parent class so traitlets work on DB.""" | |
89 | # base configurable traits: |
|
89 | # base configurable traits: | |
90 | session = Unicode("") |
|
90 | session = Unicode("") | |
91 |
|
91 | |||
92 | class DictDB(BaseDB): |
|
92 | class DictDB(BaseDB): | |
93 | """Basic in-memory dict-based object for saving Task Records. |
|
93 | """Basic in-memory dict-based object for saving Task Records. | |
94 |
|
94 | |||
95 | This is the first object to present the DB interface |
|
95 | This is the first object to present the DB interface | |
96 | for logging tasks out of memory. |
|
96 | for logging tasks out of memory. | |
97 |
|
97 | |||
98 | The interface is based on MongoDB, so adding a MongoDB |
|
98 | The interface is based on MongoDB, so adding a MongoDB | |
99 | backend should be straightforward. |
|
99 | backend should be straightforward. | |
100 | """ |
|
100 | """ | |
101 |
|
101 | |||
102 | _records = Dict() |
|
102 | _records = Dict() | |
103 |
|
103 | |||
104 | def _match_one(self, rec, tests): |
|
104 | def _match_one(self, rec, tests): | |
105 | """Check if a specific record matches tests.""" |
|
105 | """Check if a specific record matches tests.""" | |
106 | for key,test in tests.iteritems(): |
|
106 | for key,test in tests.iteritems(): | |
107 | if not test(rec.get(key, None)): |
|
107 | if not test(rec.get(key, None)): | |
108 | return False |
|
108 | return False | |
109 | return True |
|
109 | return True | |
110 |
|
110 | |||
111 | def _match(self, check): |
|
111 | def _match(self, check): | |
112 | """Find all the matches for a check dict.""" |
|
112 | """Find all the matches for a check dict.""" | |
113 | matches = [] |
|
113 | matches = [] | |
114 | tests = {} |
|
114 | tests = {} | |
115 | for k,v in check.iteritems(): |
|
115 | for k,v in check.iteritems(): | |
116 | if isinstance(v, dict): |
|
116 | if isinstance(v, dict): | |
117 | tests[k] = CompositeFilter(v) |
|
117 | tests[k] = CompositeFilter(v) | |
118 | else: |
|
118 | else: | |
119 | tests[k] = lambda o: o==v |
|
119 | tests[k] = lambda o: o==v | |
120 |
|
120 | |||
121 | for rec in self._records.itervalues(): |
|
121 | for rec in self._records.itervalues(): | |
122 | if self._match_one(rec, tests): |
|
122 | if self._match_one(rec, tests): | |
123 | matches.append(rec) |
|
123 | matches.append(rec) | |
124 | return matches |
|
124 | return matches | |
125 |
|
125 | |||
126 | def _extract_subdict(self, rec, keys): |
|
126 | def _extract_subdict(self, rec, keys): | |
127 | """extract subdict of keys""" |
|
127 | """extract subdict of keys""" | |
128 | d = {} |
|
128 | d = {} | |
129 | d['msg_id'] = rec['msg_id'] |
|
129 | d['msg_id'] = rec['msg_id'] | |
130 | for key in keys: |
|
130 | for key in keys: | |
131 | d[key] = rec[key] |
|
131 | d[key] = rec[key] | |
132 | return d |
|
132 | return d | |
133 |
|
133 | |||
134 | def add_record(self, msg_id, rec): |
|
134 | def add_record(self, msg_id, rec): | |
135 | """Add a new Task Record, by msg_id.""" |
|
135 | """Add a new Task Record, by msg_id.""" | |
136 | if self._records.has_key(msg_id): |
|
136 | if self._records.has_key(msg_id): | |
137 | raise KeyError("Already have msg_id %r"%(msg_id)) |
|
137 | raise KeyError("Already have msg_id %r"%(msg_id)) | |
138 | self._records[msg_id] = rec |
|
138 | self._records[msg_id] = rec | |
139 |
|
139 | |||
140 | def get_record(self, msg_id): |
|
140 | def get_record(self, msg_id): | |
141 | """Get a specific Task Record, by msg_id.""" |
|
141 | """Get a specific Task Record, by msg_id.""" | |
142 | if not self._records.has_key(msg_id): |
|
142 | if not self._records.has_key(msg_id): | |
143 | raise KeyError("No such msg_id %r"%(msg_id)) |
|
143 | raise KeyError("No such msg_id %r"%(msg_id)) | |
144 | return self._records[msg_id] |
|
144 | return self._records[msg_id] | |
145 |
|
145 | |||
146 | def update_record(self, msg_id, rec): |
|
146 | def update_record(self, msg_id, rec): | |
147 | """Update the data in an existing record.""" |
|
147 | """Update the data in an existing record.""" | |
148 | self._records[msg_id].update(rec) |
|
148 | self._records[msg_id].update(rec) | |
149 |
|
149 | |||
150 | def drop_matching_records(self, check): |
|
150 | def drop_matching_records(self, check): | |
151 | """Remove a record from the DB.""" |
|
151 | """Remove a record from the DB.""" | |
152 | matches = self._match(check) |
|
152 | matches = self._match(check) | |
153 | for m in matches: |
|
153 | for m in matches: | |
154 | del self._records[m['msg_id']] |
|
154 | del self._records[m['msg_id']] | |
155 |
|
155 | |||
156 | def drop_record(self, msg_id): |
|
156 | def drop_record(self, msg_id): | |
157 | """Remove a record from the DB.""" |
|
157 | """Remove a record from the DB.""" | |
158 | del self._records[msg_id] |
|
158 | del self._records[msg_id] | |
159 |
|
159 | |||
160 |
|
160 | |||
161 | def find_records(self, check, keys=None): |
|
161 | def find_records(self, check, keys=None): | |
162 | """Find records matching a query dict, optionally extracting subset of keys. |
|
162 | """Find records matching a query dict, optionally extracting subset of keys. | |
163 |
|
163 | |||
164 | Returns dict keyed by msg_id of matching records. |
|
164 | Returns dict keyed by msg_id of matching records. | |
165 |
|
165 | |||
166 | Parameters |
|
166 | Parameters | |
167 | ---------- |
|
167 | ---------- | |
168 |
|
168 | |||
169 | check: dict |
|
169 | check: dict | |
170 | mongodb-style query argument |
|
170 | mongodb-style query argument | |
171 | keys: list of strs [optional] |
|
171 | keys: list of strs [optional] | |
172 | if specified, the subset of keys to extract. msg_id will *always* be |
|
172 | if specified, the subset of keys to extract. msg_id will *always* be | |
173 | included. |
|
173 | included. | |
174 | """ |
|
174 | """ | |
175 | matches = self._match(check) |
|
175 | matches = self._match(check) | |
176 | if keys: |
|
176 | if keys: | |
177 | return [ self._extract_subdict(rec, keys) for rec in matches ] |
|
177 | return [ self._extract_subdict(rec, keys) for rec in matches ] | |
178 | else: |
|
178 | else: | |
179 | return matches |
|
179 | return matches | |
180 |
|
180 | |||
181 |
|
181 | |||
182 | def get_history(self): |
|
182 | def get_history(self): | |
183 | """get all msg_ids, ordered by time submitted.""" |
|
183 | """get all msg_ids, ordered by time submitted.""" | |
184 | msg_ids = self._records.keys() |
|
184 | msg_ids = self._records.keys() | |
185 | return sorted(msg_ids, key=lambda m: self._records[m]['submitted']) |
|
185 | return sorted(msg_ids, key=lambda m: self._records[m]['submitted']) | |
|
186 | ||||
|
187 | class NoDB(DictDB): | |||
|
188 | """A blackhole db backend that actually stores no information. | |||
|
189 | ||||
|
190 | Provides the full DB interface, but raises KeyErrors on any | |||
|
191 | method that tries to access the records. This can be used to | |||
|
192 | minimize the memory footprint of the Hub when its record-keeping | |||
|
193 | functionality is not required. | |||
|
194 | """ | |||
|
195 | ||||
|
196 | def add_record(self, msg_id, record): | |||
|
197 | pass | |||
|
198 | ||||
|
199 | def get_record(self, msg_id): | |||
|
200 | raise KeyError("NoDB does not support record access") | |||
|
201 | ||||
|
202 | def update_record(self, msg_id, record): | |||
|
203 | pass | |||
|
204 | ||||
|
205 | def drop_matching_records(self, check): | |||
|
206 | pass | |||
|
207 | ||||
|
208 | def drop_record(self, msg_id): | |||
|
209 | pass | |||
|
210 | ||||
|
211 | def find_records(self, check, keys=None): | |||
|
212 | raise KeyError("NoDB does not store information") | |||
|
213 | ||||
|
214 | def get_history(self): | |||
|
215 | raise KeyError("NoDB does not store information") | |||
|
216 |
@@ -1,114 +1,137 b'' | |||||
1 | .. _parallel_db: |
|
1 | .. _parallel_db: | |
2 |
|
2 | |||
3 | ======================= |
|
3 | ======================= | |
4 | IPython's Task Database |
|
4 | IPython's Task Database | |
5 | ======================= |
|
5 | ======================= | |
6 |
|
6 | |||
7 | The IPython Hub stores all task requests and results in a database. Currently supported backends |
|
7 | The IPython Hub stores all task requests and results in a database. Currently supported backends | |
8 | are: MongoDB, SQLite (the default), and an in-memory DictDB. The most common use case for |
|
8 | are: MongoDB, SQLite (the default), and an in-memory DictDB. The most common use case for | |
9 | this is clients requesting results for tasks they did not submit, via: |
|
9 | this is clients requesting results for tasks they did not submit, via: | |
10 |
|
10 | |||
11 | .. sourcecode:: ipython |
|
11 | .. sourcecode:: ipython | |
12 |
|
12 | |||
13 | In [1]: rc.get_result(task_id) |
|
13 | In [1]: rc.get_result(task_id) | |
14 |
|
14 | |||
15 | However, since we have this DB backend, we provide a direct query method in the :class:`client` |
|
15 | However, since we have this DB backend, we provide a direct query method in the :class:`client` | |
16 | for users who want deeper introspection into their task history. The :meth:`db_query` method of |
|
16 | for users who want deeper introspection into their task history. The :meth:`db_query` method of | |
17 | the Client is modeled after MongoDB queries, so if you have used MongoDB it should look |
|
17 | the Client is modeled after MongoDB queries, so if you have used MongoDB it should look | |
18 | familiar. In fact, when the MongoDB backend is in use, the query is relayed directly. However, |
|
18 | familiar. In fact, when the MongoDB backend is in use, the query is relayed directly. However, | |
19 | when using other backends, the interface is emulated and only a subset of queries is possible. |
|
19 | when using other backends, the interface is emulated and only a subset of queries is possible. | |
20 |
|
20 | |||
21 | .. seealso:: |
|
21 | .. seealso:: | |
22 |
|
22 | |||
23 | MongoDB query docs: http://www.mongodb.org/display/DOCS/Querying |
|
23 | MongoDB query docs: http://www.mongodb.org/display/DOCS/Querying | |
24 |
|
24 | |||
25 | :meth:`Client.db_query` takes a dictionary query object, with keys from the TaskRecord key list, |
|
25 | :meth:`Client.db_query` takes a dictionary query object, with keys from the TaskRecord key list, | |
26 | and values of either exact values to test, or MongoDB queries, which are dicts of The form: |
|
26 | and values of either exact values to test, or MongoDB queries, which are dicts of The form: | |
27 | ``{'operator' : 'argument(s)'}``. There is also an optional `keys` argument, that specifies |
|
27 | ``{'operator' : 'argument(s)'}``. There is also an optional `keys` argument, that specifies | |
28 | which subset of keys should be retrieved. The default is to retrieve all keys excluding the |
|
28 | which subset of keys should be retrieved. The default is to retrieve all keys excluding the | |
29 | request and result buffers. :meth:`db_query` returns a list of TaskRecord dicts. Also like |
|
29 | request and result buffers. :meth:`db_query` returns a list of TaskRecord dicts. Also like | |
30 | MongoDB, the `msg_id` key will always be included, whether requested or not. |
|
30 | MongoDB, the `msg_id` key will always be included, whether requested or not. | |
31 |
|
31 | |||
32 | TaskRecord keys: |
|
32 | TaskRecord keys: | |
33 |
|
33 | |||
34 | =============== =============== ============= |
|
34 | =============== =============== ============= | |
35 | Key Type Description |
|
35 | Key Type Description | |
36 | =============== =============== ============= |
|
36 | =============== =============== ============= | |
37 | msg_id uuid(bytes) The msg ID |
|
37 | msg_id uuid(bytes) The msg ID | |
38 | header dict The request header |
|
38 | header dict The request header | |
39 | content dict The request content (likely empty) |
|
39 | content dict The request content (likely empty) | |
40 | buffers list(bytes) buffers containing serialized request objects |
|
40 | buffers list(bytes) buffers containing serialized request objects | |
41 | submitted datetime timestamp for time of submission (set by client) |
|
41 | submitted datetime timestamp for time of submission (set by client) | |
42 | client_uuid uuid(bytes) IDENT of client's socket |
|
42 | client_uuid uuid(bytes) IDENT of client's socket | |
43 | engine_uuid uuid(bytes) IDENT of engine's socket |
|
43 | engine_uuid uuid(bytes) IDENT of engine's socket | |
44 | started datetime time task began execution on engine |
|
44 | started datetime time task began execution on engine | |
45 | completed datetime time task finished execution (success or failure) on engine |
|
45 | completed datetime time task finished execution (success or failure) on engine | |
46 | resubmitted datetime time of resubmission (if applicable) |
|
46 | resubmitted datetime time of resubmission (if applicable) | |
47 | result_header dict header for result |
|
47 | result_header dict header for result | |
48 | result_content dict content for result |
|
48 | result_content dict content for result | |
49 | result_buffers list(bytes) buffers containing serialized request objects |
|
49 | result_buffers list(bytes) buffers containing serialized request objects | |
50 | queue bytes The name of the queue for the task ('mux' or 'task') |
|
50 | queue bytes The name of the queue for the task ('mux' or 'task') | |
51 | pyin <unused> Python input (unused) |
|
51 | pyin <unused> Python input (unused) | |
52 | pyout <unused> Python output (unused) |
|
52 | pyout <unused> Python output (unused) | |
53 | pyerr <unused> Python traceback (unused) |
|
53 | pyerr <unused> Python traceback (unused) | |
54 | stdout str Stream of stdout data |
|
54 | stdout str Stream of stdout data | |
55 | stderr str Stream of stderr data |
|
55 | stderr str Stream of stderr data | |
56 |
|
56 | |||
57 | =============== =============== ============= |
|
57 | =============== =============== ============= | |
58 |
|
58 | |||
59 | MongoDB operators we emulate on all backends: |
|
59 | MongoDB operators we emulate on all backends: | |
60 |
|
60 | |||
61 | ========== ================= |
|
61 | ========== ================= | |
62 | Operator Python equivalent |
|
62 | Operator Python equivalent | |
63 | ========== ================= |
|
63 | ========== ================= | |
64 | '$in' in |
|
64 | '$in' in | |
65 | '$nin' not in |
|
65 | '$nin' not in | |
66 | '$eq' == |
|
66 | '$eq' == | |
67 | '$ne' != |
|
67 | '$ne' != | |
68 | '$ge' > |
|
68 | '$ge' > | |
69 | '$gte' >= |
|
69 | '$gte' >= | |
70 | '$le' < |
|
70 | '$le' < | |
71 | '$lte' <= |
|
71 | '$lte' <= | |
72 | ========== ================= |
|
72 | ========== ================= | |
73 |
|
73 | |||
74 |
|
74 | |||
75 | The DB Query is useful for two primary cases: |
|
75 | The DB Query is useful for two primary cases: | |
76 |
|
76 | |||
77 | 1. deep polling of task status or metadata |
|
77 | 1. deep polling of task status or metadata | |
78 | 2. selecting a subset of tasks, on which to perform a later operation (e.g. wait on result, purge records, resubmit,...) |
|
78 | 2. selecting a subset of tasks, on which to perform a later operation (e.g. wait on result, purge records, resubmit,...) | |
79 |
|
79 | |||
80 | Example Queries |
|
80 | Example Queries | |
81 | =============== |
|
81 | =============== | |
82 |
|
82 | |||
83 |
|
83 | |||
84 | To get all msg_ids that are not completed, only retrieving their ID and start time: |
|
84 | To get all msg_ids that are not completed, only retrieving their ID and start time: | |
85 |
|
85 | |||
86 | .. sourcecode:: ipython |
|
86 | .. sourcecode:: ipython | |
87 |
|
87 | |||
88 | In [1]: incomplete = rc.db_query({'complete' : None}, keys=['msg_id', 'started']) |
|
88 | In [1]: incomplete = rc.db_query({'complete' : None}, keys=['msg_id', 'started']) | |
89 |
|
89 | |||
90 | All jobs started in the last hour by me: |
|
90 | All jobs started in the last hour by me: | |
91 |
|
91 | |||
92 | .. sourcecode:: ipython |
|
92 | .. sourcecode:: ipython | |
93 |
|
93 | |||
94 | In [1]: from datetime import datetime, timedelta |
|
94 | In [1]: from datetime import datetime, timedelta | |
95 |
|
95 | |||
96 | In [2]: hourago = datetime.now() - timedelta(1./24) |
|
96 | In [2]: hourago = datetime.now() - timedelta(1./24) | |
97 |
|
97 | |||
98 | In [3]: recent = rc.db_query({'started' : {'$gte' : hourago }, |
|
98 | In [3]: recent = rc.db_query({'started' : {'$gte' : hourago }, | |
99 | 'client_uuid' : rc.session.session}) |
|
99 | 'client_uuid' : rc.session.session}) | |
100 |
|
100 | |||
101 | All jobs started more than an hour ago, by clients *other than me*: |
|
101 | All jobs started more than an hour ago, by clients *other than me*: | |
102 |
|
102 | |||
103 | .. sourcecode:: ipython |
|
103 | .. sourcecode:: ipython | |
104 |
|
104 | |||
105 | In [3]: recent = rc.db_query({'started' : {'$le' : hourago }, |
|
105 | In [3]: recent = rc.db_query({'started' : {'$le' : hourago }, | |
106 | 'client_uuid' : {'$ne' : rc.session.session}}) |
|
106 | 'client_uuid' : {'$ne' : rc.session.session}}) | |
107 |
|
107 | |||
108 | Result headers for all jobs on engine 3 or 4: |
|
108 | Result headers for all jobs on engine 3 or 4: | |
109 |
|
109 | |||
110 | .. sourcecode:: ipython |
|
110 | .. sourcecode:: ipython | |
111 |
|
111 | |||
112 | In [1]: uuids = map(rc._engines.get, (3,4)) |
|
112 | In [1]: uuids = map(rc._engines.get, (3,4)) | |
113 |
|
113 | |||
114 | In [2]: hist34 = rc.db_query({'engine_uuid' : {'$in' : uuids }, keys='result_header') |
|
114 | In [2]: hist34 = rc.db_query({'engine_uuid' : {'$in' : uuids }, keys='result_header') | |
|
115 | ||||
|
116 | ||||
|
117 | Cost | |||
|
118 | ==== | |||
|
119 | ||||
|
120 | The advantage of the database backends is, of course, that large amounts of | |||
|
121 | data can be stored that won't fit in memory. The default 'backend' is actually | |||
|
122 | to just store all of this information in a Python dictionary. This is very fast, | |||
|
123 | but will run out of memory quickly if you move a lot of data around, or your | |||
|
124 | cluster is to run for a long time. | |||
|
125 | ||||
|
126 | Unfortunately, the DB backends (SQLite and MongoDB) right now are rather slow, | |||
|
127 | and can still consume large amounts of resources, particularly if large tasks | |||
|
128 | or results are being created at a high frequency. | |||
|
129 | ||||
|
130 | For this reason, we have added :class:`~.NoDB`,a dummy backend that doesn't | |||
|
131 | actually store any information. When you use this database, nothing is stored, | |||
|
132 | and any request for results will result in a KeyError. This obviously prevents | |||
|
133 | later requests for results and task resubmission from functioning, but | |||
|
134 | sometimes those nice features are not as useful as keeping Hub memory under | |||
|
135 | control. | |||
|
136 | ||||
|
137 |
@@ -1,838 +1,858 b'' | |||||
1 | .. _parallel_process: |
|
1 | .. _parallel_process: | |
2 |
|
2 | |||
3 | =========================================== |
|
3 | =========================================== | |
4 | Starting the IPython controller and engines |
|
4 | Starting the IPython controller and engines | |
5 | =========================================== |
|
5 | =========================================== | |
6 |
|
6 | |||
7 | To use IPython for parallel computing, you need to start one instance of |
|
7 | To use IPython for parallel computing, you need to start one instance of | |
8 | the controller and one or more instances of the engine. The controller |
|
8 | the controller and one or more instances of the engine. The controller | |
9 | and each engine can run on different machines or on the same machine. |
|
9 | and each engine can run on different machines or on the same machine. | |
10 | Because of this, there are many different possibilities. |
|
10 | Because of this, there are many different possibilities. | |
11 |
|
11 | |||
12 | Broadly speaking, there are two ways of going about starting a controller and engines: |
|
12 | Broadly speaking, there are two ways of going about starting a controller and engines: | |
13 |
|
13 | |||
14 | * In an automated manner using the :command:`ipcluster` command. |
|
14 | * In an automated manner using the :command:`ipcluster` command. | |
15 | * In a more manual way using the :command:`ipcontroller` and |
|
15 | * In a more manual way using the :command:`ipcontroller` and | |
16 | :command:`ipengine` commands. |
|
16 | :command:`ipengine` commands. | |
17 |
|
17 | |||
18 | This document describes both of these methods. We recommend that new users |
|
18 | This document describes both of these methods. We recommend that new users | |
19 | start with the :command:`ipcluster` command as it simplifies many common usage |
|
19 | start with the :command:`ipcluster` command as it simplifies many common usage | |
20 | cases. |
|
20 | cases. | |
21 |
|
21 | |||
22 | General considerations |
|
22 | General considerations | |
23 | ====================== |
|
23 | ====================== | |
24 |
|
24 | |||
25 | Before delving into the details about how you can start a controller and |
|
25 | Before delving into the details about how you can start a controller and | |
26 | engines using the various methods, we outline some of the general issues that |
|
26 | engines using the various methods, we outline some of the general issues that | |
27 | come up when starting the controller and engines. These things come up no |
|
27 | come up when starting the controller and engines. These things come up no | |
28 | matter which method you use to start your IPython cluster. |
|
28 | matter which method you use to start your IPython cluster. | |
29 |
|
29 | |||
30 | If you are running engines on multiple machines, you will likely need to instruct the |
|
30 | If you are running engines on multiple machines, you will likely need to instruct the | |
31 | controller to listen for connections on an external interface. This can be done by specifying |
|
31 | controller to listen for connections on an external interface. This can be done by specifying | |
32 | the ``ip`` argument on the command-line, or the ``HubFactory.ip`` configurable in |
|
32 | the ``ip`` argument on the command-line, or the ``HubFactory.ip`` configurable in | |
33 | :file:`ipcontroller_config.py`. |
|
33 | :file:`ipcontroller_config.py`. | |
34 |
|
34 | |||
35 | If your machines are on a trusted network, you can safely instruct the controller to listen |
|
35 | If your machines are on a trusted network, you can safely instruct the controller to listen | |
36 | on all public interfaces with:: |
|
36 | on all public interfaces with:: | |
37 |
|
37 | |||
38 | $> ipcontroller --ip=* |
|
38 | $> ipcontroller --ip=* | |
39 |
|
39 | |||
40 | Or you can set the same behavior as the default by adding the following line to your :file:`ipcontroller_config.py`: |
|
40 | Or you can set the same behavior as the default by adding the following line to your :file:`ipcontroller_config.py`: | |
41 |
|
41 | |||
42 | .. sourcecode:: python |
|
42 | .. sourcecode:: python | |
43 |
|
43 | |||
44 | c.HubFactory.ip = '*' |
|
44 | c.HubFactory.ip = '*' | |
45 |
|
45 | |||
46 | .. note:: |
|
46 | .. note:: | |
47 |
|
47 | |||
48 | Due to the lack of security in ZeroMQ, the controller will only listen for connections on |
|
48 | Due to the lack of security in ZeroMQ, the controller will only listen for connections on | |
49 | localhost by default. If you see Timeout errors on engines or clients, then the first |
|
49 | localhost by default. If you see Timeout errors on engines or clients, then the first | |
50 | thing you should check is the ip address the controller is listening on, and make sure |
|
50 | thing you should check is the ip address the controller is listening on, and make sure | |
51 | that it is visible from the timing out machine. |
|
51 | that it is visible from the timing out machine. | |
52 |
|
52 | |||
53 | .. seealso:: |
|
53 | .. seealso:: | |
54 |
|
54 | |||
55 | Our `notes <parallel_security>`_ on security in the new parallel computing code. |
|
55 | Our `notes <parallel_security>`_ on security in the new parallel computing code. | |
56 |
|
56 | |||
57 | Let's say that you want to start the controller on ``host0`` and engines on |
|
57 | Let's say that you want to start the controller on ``host0`` and engines on | |
58 | hosts ``host1``-``hostn``. The following steps are then required: |
|
58 | hosts ``host1``-``hostn``. The following steps are then required: | |
59 |
|
59 | |||
60 | 1. Start the controller on ``host0`` by running :command:`ipcontroller` on |
|
60 | 1. Start the controller on ``host0`` by running :command:`ipcontroller` on | |
61 | ``host0``. The controller must be instructed to listen on an interface visible |
|
61 | ``host0``. The controller must be instructed to listen on an interface visible | |
62 | to the engine machines, via the ``ip`` command-line argument or ``HubFactory.ip`` |
|
62 | to the engine machines, via the ``ip`` command-line argument or ``HubFactory.ip`` | |
63 | in :file:`ipcontroller_config.py`. |
|
63 | in :file:`ipcontroller_config.py`. | |
64 | 2. Move the JSON file (:file:`ipcontroller-engine.json`) created by the |
|
64 | 2. Move the JSON file (:file:`ipcontroller-engine.json`) created by the | |
65 | controller from ``host0`` to hosts ``host1``-``hostn``. |
|
65 | controller from ``host0`` to hosts ``host1``-``hostn``. | |
66 | 3. Start the engines on hosts ``host1``-``hostn`` by running |
|
66 | 3. Start the engines on hosts ``host1``-``hostn`` by running | |
67 | :command:`ipengine`. This command has to be told where the JSON file |
|
67 | :command:`ipengine`. This command has to be told where the JSON file | |
68 | (:file:`ipcontroller-engine.json`) is located. |
|
68 | (:file:`ipcontroller-engine.json`) is located. | |
69 |
|
69 | |||
70 | At this point, the controller and engines will be connected. By default, the JSON files |
|
70 | At this point, the controller and engines will be connected. By default, the JSON files | |
71 | created by the controller are put into the :file:`~/.ipython/profile_default/security` |
|
71 | created by the controller are put into the :file:`~/.ipython/profile_default/security` | |
72 | directory. If the engines share a filesystem with the controller, step 2 can be skipped as |
|
72 | directory. If the engines share a filesystem with the controller, step 2 can be skipped as | |
73 | the engines will automatically look at that location. |
|
73 | the engines will automatically look at that location. | |
74 |
|
74 | |||
75 | The final step required to actually use the running controller from a client is to move |
|
75 | The final step required to actually use the running controller from a client is to move | |
76 | the JSON file :file:`ipcontroller-client.json` from ``host0`` to any host where clients |
|
76 | the JSON file :file:`ipcontroller-client.json` from ``host0`` to any host where clients | |
77 | will be run. If these file are put into the :file:`~/.ipython/profile_default/security` |
|
77 | will be run. If these file are put into the :file:`~/.ipython/profile_default/security` | |
78 | directory of the client's host, they will be found automatically. Otherwise, the full path |
|
78 | directory of the client's host, they will be found automatically. Otherwise, the full path | |
79 | to them has to be passed to the client's constructor. |
|
79 | to them has to be passed to the client's constructor. | |
80 |
|
80 | |||
81 | Using :command:`ipcluster` |
|
81 | Using :command:`ipcluster` | |
82 | =========================== |
|
82 | =========================== | |
83 |
|
83 | |||
84 | The :command:`ipcluster` command provides a simple way of starting a |
|
84 | The :command:`ipcluster` command provides a simple way of starting a | |
85 | controller and engines in the following situations: |
|
85 | controller and engines in the following situations: | |
86 |
|
86 | |||
87 | 1. When the controller and engines are all run on localhost. This is useful |
|
87 | 1. When the controller and engines are all run on localhost. This is useful | |
88 | for testing or running on a multicore computer. |
|
88 | for testing or running on a multicore computer. | |
89 | 2. When engines are started using the :command:`mpiexec` command that comes |
|
89 | 2. When engines are started using the :command:`mpiexec` command that comes | |
90 | with most MPI [MPI]_ implementations |
|
90 | with most MPI [MPI]_ implementations | |
91 | 3. When engines are started using the PBS [PBS]_ batch system |
|
91 | 3. When engines are started using the PBS [PBS]_ batch system | |
92 | (or other `qsub` systems, such as SGE). |
|
92 | (or other `qsub` systems, such as SGE). | |
93 | 4. When the controller is started on localhost and the engines are started on |
|
93 | 4. When the controller is started on localhost and the engines are started on | |
94 | remote nodes using :command:`ssh`. |
|
94 | remote nodes using :command:`ssh`. | |
95 | 5. When engines are started using the Windows HPC Server batch system. |
|
95 | 5. When engines are started using the Windows HPC Server batch system. | |
96 |
|
96 | |||
97 | .. note:: |
|
97 | .. note:: | |
98 |
|
98 | |||
99 | Currently :command:`ipcluster` requires that the |
|
99 | Currently :command:`ipcluster` requires that the | |
100 | :file:`~/.ipython/profile_<name>/security` directory live on a shared filesystem that is |
|
100 | :file:`~/.ipython/profile_<name>/security` directory live on a shared filesystem that is | |
101 | seen by both the controller and engines. If you don't have a shared file |
|
101 | seen by both the controller and engines. If you don't have a shared file | |
102 | system you will need to use :command:`ipcontroller` and |
|
102 | system you will need to use :command:`ipcontroller` and | |
103 | :command:`ipengine` directly. |
|
103 | :command:`ipengine` directly. | |
104 |
|
104 | |||
105 | Under the hood, :command:`ipcluster` just uses :command:`ipcontroller` |
|
105 | Under the hood, :command:`ipcluster` just uses :command:`ipcontroller` | |
106 | and :command:`ipengine` to perform the steps described above. |
|
106 | and :command:`ipengine` to perform the steps described above. | |
107 |
|
107 | |||
108 | The simplest way to use ipcluster requires no configuration, and will |
|
108 | The simplest way to use ipcluster requires no configuration, and will | |
109 | launch a controller and a number of engines on the local machine. For instance, |
|
109 | launch a controller and a number of engines on the local machine. For instance, | |
110 | to start one controller and 4 engines on localhost, just do:: |
|
110 | to start one controller and 4 engines on localhost, just do:: | |
111 |
|
111 | |||
112 | $ ipcluster start -n 4 |
|
112 | $ ipcluster start -n 4 | |
113 |
|
113 | |||
114 | To see other command line options, do:: |
|
114 | To see other command line options, do:: | |
115 |
|
115 | |||
116 | $ ipcluster -h |
|
116 | $ ipcluster -h | |
117 |
|
117 | |||
118 |
|
118 | |||
119 | Configuring an IPython cluster |
|
119 | Configuring an IPython cluster | |
120 | ============================== |
|
120 | ============================== | |
121 |
|
121 | |||
122 | Cluster configurations are stored as `profiles`. You can create a new profile with:: |
|
122 | Cluster configurations are stored as `profiles`. You can create a new profile with:: | |
123 |
|
123 | |||
124 | $ ipython profile create --parallel --profile=myprofile |
|
124 | $ ipython profile create --parallel --profile=myprofile | |
125 |
|
125 | |||
126 | This will create the directory :file:`IPYTHONDIR/profile_myprofile`, and populate it |
|
126 | This will create the directory :file:`IPYTHONDIR/profile_myprofile`, and populate it | |
127 | with the default configuration files for the three IPython cluster commands. Once |
|
127 | with the default configuration files for the three IPython cluster commands. Once | |
128 | you edit those files, you can continue to call ipcluster/ipcontroller/ipengine |
|
128 | you edit those files, you can continue to call ipcluster/ipcontroller/ipengine | |
129 | with no arguments beyond ``profile=myprofile``, and any configuration will be maintained. |
|
129 | with no arguments beyond ``profile=myprofile``, and any configuration will be maintained. | |
130 |
|
130 | |||
131 | There is no limit to the number of profiles you can have, so you can maintain a profile for each |
|
131 | There is no limit to the number of profiles you can have, so you can maintain a profile for each | |
132 | of your common use cases. The default profile will be used whenever the |
|
132 | of your common use cases. The default profile will be used whenever the | |
133 | profile argument is not specified, so edit :file:`IPYTHONDIR/profile_default/*_config.py` to |
|
133 | profile argument is not specified, so edit :file:`IPYTHONDIR/profile_default/*_config.py` to | |
134 | represent your most common use case. |
|
134 | represent your most common use case. | |
135 |
|
135 | |||
136 | The configuration files are loaded with commented-out settings and explanations, |
|
136 | The configuration files are loaded with commented-out settings and explanations, | |
137 | which should cover most of the available possibilities. |
|
137 | which should cover most of the available possibilities. | |
138 |
|
138 | |||
139 | Using various batch systems with :command:`ipcluster` |
|
139 | Using various batch systems with :command:`ipcluster` | |
140 | ----------------------------------------------------- |
|
140 | ----------------------------------------------------- | |
141 |
|
141 | |||
142 | :command:`ipcluster` has a notion of Launchers that can start controllers |
|
142 | :command:`ipcluster` has a notion of Launchers that can start controllers | |
143 | and engines with various remote execution schemes. Currently supported |
|
143 | and engines with various remote execution schemes. Currently supported | |
144 | models include :command:`ssh`, :command:`mpiexec`, PBS-style (Torque, SGE, LSF), |
|
144 | models include :command:`ssh`, :command:`mpiexec`, PBS-style (Torque, SGE, LSF), | |
145 | and Windows HPC Server. |
|
145 | and Windows HPC Server. | |
146 |
|
146 | |||
147 | In general, these are configured by the :attr:`IPClusterEngines.engine_set_launcher_class`, |
|
147 | In general, these are configured by the :attr:`IPClusterEngines.engine_set_launcher_class`, | |
148 | and :attr:`IPClusterStart.controller_launcher_class` configurables, which can be the |
|
148 | and :attr:`IPClusterStart.controller_launcher_class` configurables, which can be the | |
149 | fully specified object name (e.g. ``'IPython.parallel.apps.launcher.LocalControllerLauncher'``), |
|
149 | fully specified object name (e.g. ``'IPython.parallel.apps.launcher.LocalControllerLauncher'``), | |
150 | but if you are using IPython's builtin launchers, you can specify just the class name, |
|
150 | but if you are using IPython's builtin launchers, you can specify just the class name, | |
151 | or even just the prefix e.g: |
|
151 | or even just the prefix e.g: | |
152 |
|
152 | |||
153 | .. sourcecode:: python |
|
153 | .. sourcecode:: python | |
154 |
|
154 | |||
155 | c.IPClusterEngines.engine_launcher_class = 'SSH' |
|
155 | c.IPClusterEngines.engine_launcher_class = 'SSH' | |
156 | # equivalent to |
|
156 | # equivalent to | |
157 | c.IPClusterEngines.engine_launcher_class = 'SSHEngineSetLauncher' |
|
157 | c.IPClusterEngines.engine_launcher_class = 'SSHEngineSetLauncher' | |
158 | # both of which expand to |
|
158 | # both of which expand to | |
159 | c.IPClusterEngines.engine_launcher_class = 'IPython.parallel.apps.launcher.SSHEngineSetLauncher' |
|
159 | c.IPClusterEngines.engine_launcher_class = 'IPython.parallel.apps.launcher.SSHEngineSetLauncher' | |
160 |
|
160 | |||
161 | The shortest form being of particular use on the command line, where all you need to do to |
|
161 | The shortest form being of particular use on the command line, where all you need to do to | |
162 | get an IPython cluster running with engines started with MPI is: |
|
162 | get an IPython cluster running with engines started with MPI is: | |
163 |
|
163 | |||
164 | .. sourcecode:: bash |
|
164 | .. sourcecode:: bash | |
165 |
|
165 | |||
166 | $> ipcluster start --engines=MPI |
|
166 | $> ipcluster start --engines=MPI | |
167 |
|
167 | |||
168 | Assuming that the default MPI config is sufficient. |
|
168 | Assuming that the default MPI config is sufficient. | |
169 |
|
169 | |||
170 | .. note:: |
|
170 | .. note:: | |
171 |
|
171 | |||
172 | shortcuts for builtin launcher names were added in 0.12, as was the ``_class`` suffix |
|
172 | shortcuts for builtin launcher names were added in 0.12, as was the ``_class`` suffix | |
173 | on the configurable names. If you use the old 0.11 names (e.g. ``engine_set_launcher``), |
|
173 | on the configurable names. If you use the old 0.11 names (e.g. ``engine_set_launcher``), | |
174 | they will still work, but you will get a deprecation warning that the name has changed. |
|
174 | they will still work, but you will get a deprecation warning that the name has changed. | |
175 |
|
175 | |||
176 |
|
176 | |||
177 | .. note:: |
|
177 | .. note:: | |
178 |
|
178 | |||
179 | The Launchers and configuration are designed in such a way that advanced |
|
179 | The Launchers and configuration are designed in such a way that advanced | |
180 | users can subclass and configure them to fit their own system that we |
|
180 | users can subclass and configure them to fit their own system that we | |
181 | have not yet supported (such as Condor) |
|
181 | have not yet supported (such as Condor) | |
182 |
|
182 | |||
183 | Using :command:`ipcluster` in mpiexec/mpirun mode |
|
183 | Using :command:`ipcluster` in mpiexec/mpirun mode | |
184 | ------------------------------------------------- |
|
184 | ------------------------------------------------- | |
185 |
|
185 | |||
186 |
|
186 | |||
187 | The mpiexec/mpirun mode is useful if you: |
|
187 | The mpiexec/mpirun mode is useful if you: | |
188 |
|
188 | |||
189 | 1. Have MPI installed. |
|
189 | 1. Have MPI installed. | |
190 | 2. Your systems are configured to use the :command:`mpiexec` or |
|
190 | 2. Your systems are configured to use the :command:`mpiexec` or | |
191 | :command:`mpirun` commands to start MPI processes. |
|
191 | :command:`mpirun` commands to start MPI processes. | |
192 |
|
192 | |||
193 | If these are satisfied, you can create a new profile:: |
|
193 | If these are satisfied, you can create a new profile:: | |
194 |
|
194 | |||
195 | $ ipython profile create --parallel --profile=mpi |
|
195 | $ ipython profile create --parallel --profile=mpi | |
196 |
|
196 | |||
197 | and edit the file :file:`IPYTHONDIR/profile_mpi/ipcluster_config.py`. |
|
197 | and edit the file :file:`IPYTHONDIR/profile_mpi/ipcluster_config.py`. | |
198 |
|
198 | |||
199 | There, instruct ipcluster to use the MPI launchers by adding the lines: |
|
199 | There, instruct ipcluster to use the MPI launchers by adding the lines: | |
200 |
|
200 | |||
201 | .. sourcecode:: python |
|
201 | .. sourcecode:: python | |
202 |
|
202 | |||
203 | c.IPClusterEngines.engine_launcher_class = 'MPIEngineSetLauncher' |
|
203 | c.IPClusterEngines.engine_launcher_class = 'MPIEngineSetLauncher' | |
204 |
|
204 | |||
205 | If the default MPI configuration is correct, then you can now start your cluster, with:: |
|
205 | If the default MPI configuration is correct, then you can now start your cluster, with:: | |
206 |
|
206 | |||
207 | $ ipcluster start -n 4 --profile=mpi |
|
207 | $ ipcluster start -n 4 --profile=mpi | |
208 |
|
208 | |||
209 | This does the following: |
|
209 | This does the following: | |
210 |
|
210 | |||
211 | 1. Starts the IPython controller on current host. |
|
211 | 1. Starts the IPython controller on current host. | |
212 | 2. Uses :command:`mpiexec` to start 4 engines. |
|
212 | 2. Uses :command:`mpiexec` to start 4 engines. | |
213 |
|
213 | |||
214 | If you have a reason to also start the Controller with mpi, you can specify: |
|
214 | If you have a reason to also start the Controller with mpi, you can specify: | |
215 |
|
215 | |||
216 | .. sourcecode:: python |
|
216 | .. sourcecode:: python | |
217 |
|
217 | |||
218 | c.IPClusterStart.controller_launcher_class = 'MPIControllerLauncher' |
|
218 | c.IPClusterStart.controller_launcher_class = 'MPIControllerLauncher' | |
219 |
|
219 | |||
220 | .. note:: |
|
220 | .. note:: | |
221 |
|
221 | |||
222 | The Controller *will not* be in the same MPI universe as the engines, so there is not |
|
222 | The Controller *will not* be in the same MPI universe as the engines, so there is not | |
223 | much reason to do this unless sysadmins demand it. |
|
223 | much reason to do this unless sysadmins demand it. | |
224 |
|
224 | |||
225 | On newer MPI implementations (such as OpenMPI), this will work even if you |
|
225 | On newer MPI implementations (such as OpenMPI), this will work even if you | |
226 | don't make any calls to MPI or call :func:`MPI_Init`. However, older MPI |
|
226 | don't make any calls to MPI or call :func:`MPI_Init`. However, older MPI | |
227 | implementations actually require each process to call :func:`MPI_Init` upon |
|
227 | implementations actually require each process to call :func:`MPI_Init` upon | |
228 | starting. The easiest way of having this done is to install the mpi4py |
|
228 | starting. The easiest way of having this done is to install the mpi4py | |
229 | [mpi4py]_ package and then specify the ``c.MPI.use`` option in :file:`ipengine_config.py`: |
|
229 | [mpi4py]_ package and then specify the ``c.MPI.use`` option in :file:`ipengine_config.py`: | |
230 |
|
230 | |||
231 | .. sourcecode:: python |
|
231 | .. sourcecode:: python | |
232 |
|
232 | |||
233 | c.MPI.use = 'mpi4py' |
|
233 | c.MPI.use = 'mpi4py' | |
234 |
|
234 | |||
235 | Unfortunately, even this won't work for some MPI implementations. If you are |
|
235 | Unfortunately, even this won't work for some MPI implementations. If you are | |
236 | having problems with this, you will likely have to use a custom Python |
|
236 | having problems with this, you will likely have to use a custom Python | |
237 | executable that itself calls :func:`MPI_Init` at the appropriate time. |
|
237 | executable that itself calls :func:`MPI_Init` at the appropriate time. | |
238 | Fortunately, mpi4py comes with such a custom Python executable that is easy to |
|
238 | Fortunately, mpi4py comes with such a custom Python executable that is easy to | |
239 | install and use. However, this custom Python executable approach will not work |
|
239 | install and use. However, this custom Python executable approach will not work | |
240 | with :command:`ipcluster` currently. |
|
240 | with :command:`ipcluster` currently. | |
241 |
|
241 | |||
242 | More details on using MPI with IPython can be found :ref:`here <parallelmpi>`. |
|
242 | More details on using MPI with IPython can be found :ref:`here <parallelmpi>`. | |
243 |
|
243 | |||
244 |
|
244 | |||
245 | Using :command:`ipcluster` in PBS mode |
|
245 | Using :command:`ipcluster` in PBS mode | |
246 | -------------------------------------- |
|
246 | -------------------------------------- | |
247 |
|
247 | |||
248 | The PBS mode uses the Portable Batch System (PBS) to start the engines. |
|
248 | The PBS mode uses the Portable Batch System (PBS) to start the engines. | |
249 |
|
249 | |||
250 | As usual, we will start by creating a fresh profile:: |
|
250 | As usual, we will start by creating a fresh profile:: | |
251 |
|
251 | |||
252 | $ ipython profile create --parallel --profile=pbs |
|
252 | $ ipython profile create --parallel --profile=pbs | |
253 |
|
253 | |||
254 | And in :file:`ipcluster_config.py`, we will select the PBS launchers for the controller |
|
254 | And in :file:`ipcluster_config.py`, we will select the PBS launchers for the controller | |
255 | and engines: |
|
255 | and engines: | |
256 |
|
256 | |||
257 | .. sourcecode:: python |
|
257 | .. sourcecode:: python | |
258 |
|
258 | |||
259 | c.IPClusterStart.controller_launcher_class = 'PBSControllerLauncher' |
|
259 | c.IPClusterStart.controller_launcher_class = 'PBSControllerLauncher' | |
260 | c.IPClusterEngines.engine_launcher_class = 'PBSEngineSetLauncher' |
|
260 | c.IPClusterEngines.engine_launcher_class = 'PBSEngineSetLauncher' | |
261 |
|
261 | |||
262 | .. note:: |
|
262 | .. note:: | |
263 |
|
263 | |||
264 | Note that the configurable is IPClusterEngines for the engine launcher, and |
|
264 | Note that the configurable is IPClusterEngines for the engine launcher, and | |
265 | IPClusterStart for the controller launcher. This is because the start command is a |
|
265 | IPClusterStart for the controller launcher. This is because the start command is a | |
266 | subclass of the engine command, adding a controller launcher. Since it is a subclass, |
|
266 | subclass of the engine command, adding a controller launcher. Since it is a subclass, | |
267 | any configuration made in IPClusterEngines is inherited by IPClusterStart unless it is |
|
267 | any configuration made in IPClusterEngines is inherited by IPClusterStart unless it is | |
268 | overridden. |
|
268 | overridden. | |
269 |
|
269 | |||
270 | IPython does provide simple default batch templates for PBS and SGE, but you may need |
|
270 | IPython does provide simple default batch templates for PBS and SGE, but you may need | |
271 | to specify your own. Here is a sample PBS script template: |
|
271 | to specify your own. Here is a sample PBS script template: | |
272 |
|
272 | |||
273 | .. sourcecode:: bash |
|
273 | .. sourcecode:: bash | |
274 |
|
274 | |||
275 | #PBS -N ipython |
|
275 | #PBS -N ipython | |
276 | #PBS -j oe |
|
276 | #PBS -j oe | |
277 | #PBS -l walltime=00:10:00 |
|
277 | #PBS -l walltime=00:10:00 | |
278 | #PBS -l nodes={n/4}:ppn=4 |
|
278 | #PBS -l nodes={n/4}:ppn=4 | |
279 | #PBS -q {queue} |
|
279 | #PBS -q {queue} | |
280 |
|
280 | |||
281 | cd $PBS_O_WORKDIR |
|
281 | cd $PBS_O_WORKDIR | |
282 | export PATH=$HOME/usr/local/bin |
|
282 | export PATH=$HOME/usr/local/bin | |
283 | export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages |
|
283 | export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages | |
284 | /usr/local/bin/mpiexec -n {n} ipengine --profile-dir={profile_dir} |
|
284 | /usr/local/bin/mpiexec -n {n} ipengine --profile-dir={profile_dir} | |
285 |
|
285 | |||
286 | There are a few important points about this template: |
|
286 | There are a few important points about this template: | |
287 |
|
287 | |||
288 | 1. This template will be rendered at runtime using IPython's :class:`EvalFormatter`. |
|
288 | 1. This template will be rendered at runtime using IPython's :class:`EvalFormatter`. | |
289 | This is simply a subclass of :class:`string.Formatter` that allows simple expressions |
|
289 | This is simply a subclass of :class:`string.Formatter` that allows simple expressions | |
290 | on keys. |
|
290 | on keys. | |
291 |
|
291 | |||
292 | 2. Instead of putting in the actual number of engines, use the notation |
|
292 | 2. Instead of putting in the actual number of engines, use the notation | |
293 | ``{n}`` to indicate the number of engines to be started. You can also use |
|
293 | ``{n}`` to indicate the number of engines to be started. You can also use | |
294 | expressions like ``{n/4}`` in the template to indicate the number of nodes. |
|
294 | expressions like ``{n/4}`` in the template to indicate the number of nodes. | |
295 | There will always be ``{n}`` and ``{profile_dir}`` variables passed to the formatter. |
|
295 | There will always be ``{n}`` and ``{profile_dir}`` variables passed to the formatter. | |
296 | These allow the batch system to know how many engines, and where the configuration |
|
296 | These allow the batch system to know how many engines, and where the configuration | |
297 | files reside. The same is true for the batch queue, with the template variable |
|
297 | files reside. The same is true for the batch queue, with the template variable | |
298 | ``{queue}``. |
|
298 | ``{queue}``. | |
299 |
|
299 | |||
300 | 3. Any options to :command:`ipengine` can be given in the batch script |
|
300 | 3. Any options to :command:`ipengine` can be given in the batch script | |
301 | template, or in :file:`ipengine_config.py`. |
|
301 | template, or in :file:`ipengine_config.py`. | |
302 |
|
302 | |||
303 | 4. Depending on the configuration of you system, you may have to set |
|
303 | 4. Depending on the configuration of you system, you may have to set | |
304 | environment variables in the script template. |
|
304 | environment variables in the script template. | |
305 |
|
305 | |||
306 | The controller template should be similar, but simpler: |
|
306 | The controller template should be similar, but simpler: | |
307 |
|
307 | |||
308 | .. sourcecode:: bash |
|
308 | .. sourcecode:: bash | |
309 |
|
309 | |||
310 | #PBS -N ipython |
|
310 | #PBS -N ipython | |
311 | #PBS -j oe |
|
311 | #PBS -j oe | |
312 | #PBS -l walltime=00:10:00 |
|
312 | #PBS -l walltime=00:10:00 | |
313 | #PBS -l nodes=1:ppn=4 |
|
313 | #PBS -l nodes=1:ppn=4 | |
314 | #PBS -q {queue} |
|
314 | #PBS -q {queue} | |
315 |
|
315 | |||
316 | cd $PBS_O_WORKDIR |
|
316 | cd $PBS_O_WORKDIR | |
317 | export PATH=$HOME/usr/local/bin |
|
317 | export PATH=$HOME/usr/local/bin | |
318 | export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages |
|
318 | export PYTHONPATH=$HOME/usr/local/lib/python2.7/site-packages | |
319 | ipcontroller --profile-dir={profile_dir} |
|
319 | ipcontroller --profile-dir={profile_dir} | |
320 |
|
320 | |||
321 |
|
321 | |||
322 | Once you have created these scripts, save them with names like |
|
322 | Once you have created these scripts, save them with names like | |
323 | :file:`pbs.engine.template`. Now you can load them into the :file:`ipcluster_config` with: |
|
323 | :file:`pbs.engine.template`. Now you can load them into the :file:`ipcluster_config` with: | |
324 |
|
324 | |||
325 | .. sourcecode:: python |
|
325 | .. sourcecode:: python | |
326 |
|
326 | |||
327 | c.PBSEngineSetLauncher.batch_template_file = "pbs.engine.template" |
|
327 | c.PBSEngineSetLauncher.batch_template_file = "pbs.engine.template" | |
328 |
|
328 | |||
329 | c.PBSControllerLauncher.batch_template_file = "pbs.controller.template" |
|
329 | c.PBSControllerLauncher.batch_template_file = "pbs.controller.template" | |
330 |
|
330 | |||
331 |
|
331 | |||
332 | Alternately, you can just define the templates as strings inside :file:`ipcluster_config`. |
|
332 | Alternately, you can just define the templates as strings inside :file:`ipcluster_config`. | |
333 |
|
333 | |||
334 | Whether you are using your own templates or our defaults, the extra configurables available are |
|
334 | Whether you are using your own templates or our defaults, the extra configurables available are | |
335 | the number of engines to launch (``{n}``, and the batch system queue to which the jobs are to be |
|
335 | the number of engines to launch (``{n}``, and the batch system queue to which the jobs are to be | |
336 | submitted (``{queue}``)). These are configurables, and can be specified in |
|
336 | submitted (``{queue}``)). These are configurables, and can be specified in | |
337 | :file:`ipcluster_config`: |
|
337 | :file:`ipcluster_config`: | |
338 |
|
338 | |||
339 | .. sourcecode:: python |
|
339 | .. sourcecode:: python | |
340 |
|
340 | |||
341 | c.PBSLauncher.queue = 'veryshort.q' |
|
341 | c.PBSLauncher.queue = 'veryshort.q' | |
342 | c.IPClusterEngines.n = 64 |
|
342 | c.IPClusterEngines.n = 64 | |
343 |
|
343 | |||
344 | Note that assuming you are running PBS on a multi-node cluster, the Controller's default behavior |
|
344 | Note that assuming you are running PBS on a multi-node cluster, the Controller's default behavior | |
345 | of listening only on localhost is likely too restrictive. In this case, also assuming the |
|
345 | of listening only on localhost is likely too restrictive. In this case, also assuming the | |
346 | nodes are safely behind a firewall, you can simply instruct the Controller to listen for |
|
346 | nodes are safely behind a firewall, you can simply instruct the Controller to listen for | |
347 | connections on all its interfaces, by adding in :file:`ipcontroller_config`: |
|
347 | connections on all its interfaces, by adding in :file:`ipcontroller_config`: | |
348 |
|
348 | |||
349 | .. sourcecode:: python |
|
349 | .. sourcecode:: python | |
350 |
|
350 | |||
351 | c.HubFactory.ip = '*' |
|
351 | c.HubFactory.ip = '*' | |
352 |
|
352 | |||
353 | You can now run the cluster with:: |
|
353 | You can now run the cluster with:: | |
354 |
|
354 | |||
355 | $ ipcluster start --profile=pbs -n 128 |
|
355 | $ ipcluster start --profile=pbs -n 128 | |
356 |
|
356 | |||
357 | Additional configuration options can be found in the PBS section of :file:`ipcluster_config`. |
|
357 | Additional configuration options can be found in the PBS section of :file:`ipcluster_config`. | |
358 |
|
358 | |||
359 | .. note:: |
|
359 | .. note:: | |
360 |
|
360 | |||
361 | Due to the flexibility of configuration, the PBS launchers work with simple changes |
|
361 | Due to the flexibility of configuration, the PBS launchers work with simple changes | |
362 | to the template for other :command:`qsub`-using systems, such as Sun Grid Engine, |
|
362 | to the template for other :command:`qsub`-using systems, such as Sun Grid Engine, | |
363 | and with further configuration in similar batch systems like Condor. |
|
363 | and with further configuration in similar batch systems like Condor. | |
364 |
|
364 | |||
365 |
|
365 | |||
366 | Using :command:`ipcluster` in SSH mode |
|
366 | Using :command:`ipcluster` in SSH mode | |
367 | -------------------------------------- |
|
367 | -------------------------------------- | |
368 |
|
368 | |||
369 |
|
369 | |||
370 | The SSH mode uses :command:`ssh` to execute :command:`ipengine` on remote |
|
370 | The SSH mode uses :command:`ssh` to execute :command:`ipengine` on remote | |
371 | nodes and :command:`ipcontroller` can be run remotely as well, or on localhost. |
|
371 | nodes and :command:`ipcontroller` can be run remotely as well, or on localhost. | |
372 |
|
372 | |||
373 | .. note:: |
|
373 | .. note:: | |
374 |
|
374 | |||
375 | When using this mode it highly recommended that you have set up SSH keys |
|
375 | When using this mode it highly recommended that you have set up SSH keys | |
376 | and are using ssh-agent [SSH]_ for password-less logins. |
|
376 | and are using ssh-agent [SSH]_ for password-less logins. | |
377 |
|
377 | |||
378 | As usual, we start by creating a clean profile:: |
|
378 | As usual, we start by creating a clean profile:: | |
379 |
|
379 | |||
380 | $ ipython profile create --parallel --profile=ssh |
|
380 | $ ipython profile create --parallel --profile=ssh | |
381 |
|
381 | |||
382 | To use this mode, select the SSH launchers in :file:`ipcluster_config.py`: |
|
382 | To use this mode, select the SSH launchers in :file:`ipcluster_config.py`: | |
383 |
|
383 | |||
384 | .. sourcecode:: python |
|
384 | .. sourcecode:: python | |
385 |
|
385 | |||
386 | c.IPClusterEngines.engine_launcher_class = 'SSHEngineSetLauncher' |
|
386 | c.IPClusterEngines.engine_launcher_class = 'SSHEngineSetLauncher' | |
387 | # and if the Controller is also to be remote: |
|
387 | # and if the Controller is also to be remote: | |
388 | c.IPClusterStart.controller_launcher_class = 'SSHControllerLauncher' |
|
388 | c.IPClusterStart.controller_launcher_class = 'SSHControllerLauncher' | |
389 |
|
389 | |||
390 |
|
390 | |||
391 |
|
391 | |||
392 | The controller's remote location and configuration can be specified: |
|
392 | The controller's remote location and configuration can be specified: | |
393 |
|
393 | |||
394 | .. sourcecode:: python |
|
394 | .. sourcecode:: python | |
395 |
|
395 | |||
396 | # Set the user and hostname for the controller |
|
396 | # Set the user and hostname for the controller | |
397 | # c.SSHControllerLauncher.hostname = 'controller.example.com' |
|
397 | # c.SSHControllerLauncher.hostname = 'controller.example.com' | |
398 | # c.SSHControllerLauncher.user = os.environ.get('USER','username') |
|
398 | # c.SSHControllerLauncher.user = os.environ.get('USER','username') | |
399 |
|
399 | |||
400 | # Set the arguments to be passed to ipcontroller |
|
400 | # Set the arguments to be passed to ipcontroller | |
401 | # note that remotely launched ipcontroller will not get the contents of |
|
401 | # note that remotely launched ipcontroller will not get the contents of | |
402 | # the local ipcontroller_config.py unless it resides on the *remote host* |
|
402 | # the local ipcontroller_config.py unless it resides on the *remote host* | |
403 | # in the location specified by the `profile-dir` argument. |
|
403 | # in the location specified by the `profile-dir` argument. | |
404 | # c.SSHControllerLauncher.controller_args = ['--reuse', '--ip=*', '--profile-dir=/path/to/cd'] |
|
404 | # c.SSHControllerLauncher.controller_args = ['--reuse', '--ip=*', '--profile-dir=/path/to/cd'] | |
405 |
|
405 | |||
406 | .. note:: |
|
406 | .. note:: | |
407 |
|
407 | |||
408 | SSH mode does not do any file movement, so you will need to distribute configuration |
|
408 | SSH mode does not do any file movement, so you will need to distribute configuration | |
409 | files manually. To aid in this, the `reuse_files` flag defaults to True for ssh-launched |
|
409 | files manually. To aid in this, the `reuse_files` flag defaults to True for ssh-launched | |
410 | Controllers, so you will only need to do this once, unless you override this flag back |
|
410 | Controllers, so you will only need to do this once, unless you override this flag back | |
411 | to False. |
|
411 | to False. | |
412 |
|
412 | |||
413 | Engines are specified in a dictionary, by hostname and the number of engines to be run |
|
413 | Engines are specified in a dictionary, by hostname and the number of engines to be run | |
414 | on that host. |
|
414 | on that host. | |
415 |
|
415 | |||
416 | .. sourcecode:: python |
|
416 | .. sourcecode:: python | |
417 |
|
417 | |||
418 | c.SSHEngineSetLauncher.engines = { 'host1.example.com' : 2, |
|
418 | c.SSHEngineSetLauncher.engines = { 'host1.example.com' : 2, | |
419 | 'host2.example.com' : 5, |
|
419 | 'host2.example.com' : 5, | |
420 | 'host3.example.com' : (1, ['--profile-dir=/home/different/location']), |
|
420 | 'host3.example.com' : (1, ['--profile-dir=/home/different/location']), | |
421 | 'host4.example.com' : 8 } |
|
421 | 'host4.example.com' : 8 } | |
422 |
|
422 | |||
423 | * The `engines` dict, where the keys are the host we want to run engines on and |
|
423 | * The `engines` dict, where the keys are the host we want to run engines on and | |
424 | the value is the number of engines to run on that host. |
|
424 | the value is the number of engines to run on that host. | |
425 | * on host3, the value is a tuple, where the number of engines is first, and the arguments |
|
425 | * on host3, the value is a tuple, where the number of engines is first, and the arguments | |
426 | to be passed to :command:`ipengine` are the second element. |
|
426 | to be passed to :command:`ipengine` are the second element. | |
427 |
|
427 | |||
428 | For engines without explicitly specified arguments, the default arguments are set in |
|
428 | For engines without explicitly specified arguments, the default arguments are set in | |
429 | a single location: |
|
429 | a single location: | |
430 |
|
430 | |||
431 | .. sourcecode:: python |
|
431 | .. sourcecode:: python | |
432 |
|
432 | |||
433 | c.SSHEngineSetLauncher.engine_args = ['--profile-dir=/path/to/profile_ssh'] |
|
433 | c.SSHEngineSetLauncher.engine_args = ['--profile-dir=/path/to/profile_ssh'] | |
434 |
|
434 | |||
435 | Current limitations of the SSH mode of :command:`ipcluster` are: |
|
435 | Current limitations of the SSH mode of :command:`ipcluster` are: | |
436 |
|
436 | |||
437 | * Untested on Windows. Would require a working :command:`ssh` on Windows. |
|
437 | * Untested on Windows. Would require a working :command:`ssh` on Windows. | |
438 | Also, we are using shell scripts to setup and execute commands on remote |
|
438 | Also, we are using shell scripts to setup and execute commands on remote | |
439 | hosts. |
|
439 | hosts. | |
440 | * No file movement - This is a regression from 0.10, which moved connection files |
|
440 | * No file movement - This is a regression from 0.10, which moved connection files | |
441 | around with scp. This will be improved, Pull Requests are welcome. |
|
441 | around with scp. This will be improved, Pull Requests are welcome. | |
442 |
|
442 | |||
443 |
|
443 | |||
444 | IPython on EC2 with StarCluster |
|
444 | IPython on EC2 with StarCluster | |
445 | =============================== |
|
445 | =============================== | |
446 |
|
446 | |||
447 | The excellent StarCluster_ toolkit for managing `Amazon EC2`_ clusters has a plugin |
|
447 | The excellent StarCluster_ toolkit for managing `Amazon EC2`_ clusters has a plugin | |
448 | which makes deploying IPython on EC2 quite simple. The starcluster plugin uses |
|
448 | which makes deploying IPython on EC2 quite simple. The starcluster plugin uses | |
449 | :command:`ipcluster` with the SGE launchers to distribute engines across the |
|
449 | :command:`ipcluster` with the SGE launchers to distribute engines across the | |
450 | EC2 cluster. See their `ipcluster plugin documentation`_ for more information. |
|
450 | EC2 cluster. See their `ipcluster plugin documentation`_ for more information. | |
451 |
|
451 | |||
452 | .. _StarCluster: http://web.mit.edu/starcluster |
|
452 | .. _StarCluster: http://web.mit.edu/starcluster | |
453 | .. _Amazon EC2: http://aws.amazon.com/ec2/ |
|
453 | .. _Amazon EC2: http://aws.amazon.com/ec2/ | |
454 | .. _ipcluster plugin documentation: http://web.mit.edu/starcluster/docs/latest/plugins/ipython.html |
|
454 | .. _ipcluster plugin documentation: http://web.mit.edu/starcluster/docs/latest/plugins/ipython.html | |
455 |
|
455 | |||
456 |
|
456 | |||
457 | Using the :command:`ipcontroller` and :command:`ipengine` commands |
|
457 | Using the :command:`ipcontroller` and :command:`ipengine` commands | |
458 | ================================================================== |
|
458 | ================================================================== | |
459 |
|
459 | |||
460 | It is also possible to use the :command:`ipcontroller` and :command:`ipengine` |
|
460 | It is also possible to use the :command:`ipcontroller` and :command:`ipengine` | |
461 | commands to start your controller and engines. This approach gives you full |
|
461 | commands to start your controller and engines. This approach gives you full | |
462 | control over all aspects of the startup process. |
|
462 | control over all aspects of the startup process. | |
463 |
|
463 | |||
464 | Starting the controller and engine on your local machine |
|
464 | Starting the controller and engine on your local machine | |
465 | -------------------------------------------------------- |
|
465 | -------------------------------------------------------- | |
466 |
|
466 | |||
467 | To use :command:`ipcontroller` and :command:`ipengine` to start things on your |
|
467 | To use :command:`ipcontroller` and :command:`ipengine` to start things on your | |
468 | local machine, do the following. |
|
468 | local machine, do the following. | |
469 |
|
469 | |||
470 | First start the controller:: |
|
470 | First start the controller:: | |
471 |
|
471 | |||
472 | $ ipcontroller |
|
472 | $ ipcontroller | |
473 |
|
473 | |||
474 | Next, start however many instances of the engine you want using (repeatedly) |
|
474 | Next, start however many instances of the engine you want using (repeatedly) | |
475 | the command:: |
|
475 | the command:: | |
476 |
|
476 | |||
477 | $ ipengine |
|
477 | $ ipengine | |
478 |
|
478 | |||
479 | The engines should start and automatically connect to the controller using the |
|
479 | The engines should start and automatically connect to the controller using the | |
480 | JSON files in :file:`~/.ipython/profile_default/security`. You are now ready to use the |
|
480 | JSON files in :file:`~/.ipython/profile_default/security`. You are now ready to use the | |
481 | controller and engines from IPython. |
|
481 | controller and engines from IPython. | |
482 |
|
482 | |||
483 | .. warning:: |
|
483 | .. warning:: | |
484 |
|
484 | |||
485 | The order of the above operations may be important. You *must* |
|
485 | The order of the above operations may be important. You *must* | |
486 | start the controller before the engines, unless you are reusing connection |
|
486 | start the controller before the engines, unless you are reusing connection | |
487 | information (via ``--reuse``), in which case ordering is not important. |
|
487 | information (via ``--reuse``), in which case ordering is not important. | |
488 |
|
488 | |||
489 | .. note:: |
|
489 | .. note:: | |
490 |
|
490 | |||
491 | On some platforms (OS X), to put the controller and engine into the |
|
491 | On some platforms (OS X), to put the controller and engine into the | |
492 | background you may need to give these commands in the form ``(ipcontroller |
|
492 | background you may need to give these commands in the form ``(ipcontroller | |
493 | &)`` and ``(ipengine &)`` (with the parentheses) for them to work |
|
493 | &)`` and ``(ipengine &)`` (with the parentheses) for them to work | |
494 | properly. |
|
494 | properly. | |
495 |
|
495 | |||
496 | Starting the controller and engines on different hosts |
|
496 | Starting the controller and engines on different hosts | |
497 | ------------------------------------------------------ |
|
497 | ------------------------------------------------------ | |
498 |
|
498 | |||
499 | When the controller and engines are running on different hosts, things are |
|
499 | When the controller and engines are running on different hosts, things are | |
500 | slightly more complicated, but the underlying ideas are the same: |
|
500 | slightly more complicated, but the underlying ideas are the same: | |
501 |
|
501 | |||
502 | 1. Start the controller on a host using :command:`ipcontroller`. The controller must be |
|
502 | 1. Start the controller on a host using :command:`ipcontroller`. The controller must be | |
503 | instructed to listen on an interface visible to the engine machines, via the ``ip`` |
|
503 | instructed to listen on an interface visible to the engine machines, via the ``ip`` | |
504 | command-line argument or ``HubFactory.ip`` in :file:`ipcontroller_config.py`:: |
|
504 | command-line argument or ``HubFactory.ip`` in :file:`ipcontroller_config.py`:: | |
505 |
|
505 | |||
506 | $ ipcontroller --ip=192.168.1.16 |
|
506 | $ ipcontroller --ip=192.168.1.16 | |
507 |
|
507 | |||
508 | .. sourcecode:: python |
|
508 | .. sourcecode:: python | |
509 |
|
509 | |||
510 | # in ipcontroller_config.py |
|
510 | # in ipcontroller_config.py | |
511 | HubFactory.ip = '192.168.1.16' |
|
511 | HubFactory.ip = '192.168.1.16' | |
512 |
|
512 | |||
513 | 2. Copy :file:`ipcontroller-engine.json` from :file:`~/.ipython/profile_<name>/security` on |
|
513 | 2. Copy :file:`ipcontroller-engine.json` from :file:`~/.ipython/profile_<name>/security` on | |
514 | the controller's host to the host where the engines will run. |
|
514 | the controller's host to the host where the engines will run. | |
515 | 3. Use :command:`ipengine` on the engine's hosts to start the engines. |
|
515 | 3. Use :command:`ipengine` on the engine's hosts to start the engines. | |
516 |
|
516 | |||
517 | The only thing you have to be careful of is to tell :command:`ipengine` where |
|
517 | The only thing you have to be careful of is to tell :command:`ipengine` where | |
518 | the :file:`ipcontroller-engine.json` file is located. There are two ways you |
|
518 | the :file:`ipcontroller-engine.json` file is located. There are two ways you | |
519 | can do this: |
|
519 | can do this: | |
520 |
|
520 | |||
521 | * Put :file:`ipcontroller-engine.json` in the :file:`~/.ipython/profile_<name>/security` |
|
521 | * Put :file:`ipcontroller-engine.json` in the :file:`~/.ipython/profile_<name>/security` | |
522 | directory on the engine's host, where it will be found automatically. |
|
522 | directory on the engine's host, where it will be found automatically. | |
523 | * Call :command:`ipengine` with the ``--file=full_path_to_the_file`` |
|
523 | * Call :command:`ipengine` with the ``--file=full_path_to_the_file`` | |
524 | flag. |
|
524 | flag. | |
525 |
|
525 | |||
526 | The ``file`` flag works like this:: |
|
526 | The ``file`` flag works like this:: | |
527 |
|
527 | |||
528 | $ ipengine --file=/path/to/my/ipcontroller-engine.json |
|
528 | $ ipengine --file=/path/to/my/ipcontroller-engine.json | |
529 |
|
529 | |||
530 | .. note:: |
|
530 | .. note:: | |
531 |
|
531 | |||
532 | If the controller's and engine's hosts all have a shared file system |
|
532 | If the controller's and engine's hosts all have a shared file system | |
533 | (:file:`~/.ipython/profile_<name>/security` is the same on all of them), then things |
|
533 | (:file:`~/.ipython/profile_<name>/security` is the same on all of them), then things | |
534 | will just work! |
|
534 | will just work! | |
535 |
|
535 | |||
536 | SSH Tunnels |
|
536 | SSH Tunnels | |
537 | *********** |
|
537 | *********** | |
538 |
|
538 | |||
539 | If your engines are not on the same LAN as the controller, or you are on a highly |
|
539 | If your engines are not on the same LAN as the controller, or you are on a highly | |
540 | restricted network where your nodes cannot see each others ports, then you can |
|
540 | restricted network where your nodes cannot see each others ports, then you can | |
541 | use SSH tunnels to connect engines to the controller. |
|
541 | use SSH tunnels to connect engines to the controller. | |
542 |
|
542 | |||
543 | .. note:: |
|
543 | .. note:: | |
544 |
|
544 | |||
545 | This does not work in all cases. Manual tunnels may be an option, but are |
|
545 | This does not work in all cases. Manual tunnels may be an option, but are | |
546 | highly inconvenient. Support for manual tunnels will be improved. |
|
546 | highly inconvenient. Support for manual tunnels will be improved. | |
547 |
|
547 | |||
548 | You can instruct all engines to use ssh, by specifying the ssh server in |
|
548 | You can instruct all engines to use ssh, by specifying the ssh server in | |
549 | :file:`ipcontroller-engine.json`: |
|
549 | :file:`ipcontroller-engine.json`: | |
550 |
|
550 | |||
551 | .. I know this is really JSON, but the example is a subset of Python: |
|
551 | .. I know this is really JSON, but the example is a subset of Python: | |
552 | .. sourcecode:: python |
|
552 | .. sourcecode:: python | |
553 |
|
553 | |||
554 | { |
|
554 | { | |
555 | "url":"tcp://192.168.1.123:56951", |
|
555 | "url":"tcp://192.168.1.123:56951", | |
556 | "exec_key":"26f4c040-587d-4a4e-b58b-030b96399584", |
|
556 | "exec_key":"26f4c040-587d-4a4e-b58b-030b96399584", | |
557 | "ssh":"user@example.com", |
|
557 | "ssh":"user@example.com", | |
558 | "location":"192.168.1.123" |
|
558 | "location":"192.168.1.123" | |
559 | } |
|
559 | } | |
560 |
|
560 | |||
561 | This will be specified if you give the ``--enginessh=use@example.com`` argument when |
|
561 | This will be specified if you give the ``--enginessh=use@example.com`` argument when | |
562 | starting :command:`ipcontroller`. |
|
562 | starting :command:`ipcontroller`. | |
563 |
|
563 | |||
564 | Or you can specify an ssh server on the command-line when starting an engine:: |
|
564 | Or you can specify an ssh server on the command-line when starting an engine:: | |
565 |
|
565 | |||
566 | $> ipengine --profile=foo --ssh=my.login.node |
|
566 | $> ipengine --profile=foo --ssh=my.login.node | |
567 |
|
567 | |||
568 | For example, if your system is totally restricted, then all connections will actually be |
|
568 | For example, if your system is totally restricted, then all connections will actually be | |
569 | loopback, and ssh tunnels will be used to connect engines to the controller:: |
|
569 | loopback, and ssh tunnels will be used to connect engines to the controller:: | |
570 |
|
570 | |||
571 | [node1] $> ipcontroller --enginessh=node1 |
|
571 | [node1] $> ipcontroller --enginessh=node1 | |
572 | [node2] $> ipengine |
|
572 | [node2] $> ipengine | |
573 | [node3] $> ipcluster engines --n=4 |
|
573 | [node3] $> ipcluster engines --n=4 | |
574 |
|
574 | |||
575 | Or if you want to start many engines on each node, the command `ipcluster engines --n=4` |
|
575 | Or if you want to start many engines on each node, the command `ipcluster engines --n=4` | |
576 | without any configuration is equivalent to running ipengine 4 times. |
|
576 | without any configuration is equivalent to running ipengine 4 times. | |
577 |
|
577 | |||
578 | An example using ipcontroller/engine with ssh |
|
578 | An example using ipcontroller/engine with ssh | |
579 | --------------------------------------------- |
|
579 | --------------------------------------------- | |
580 |
|
580 | |||
581 | No configuration files are necessary to use ipcontroller/engine in an SSH environment |
|
581 | No configuration files are necessary to use ipcontroller/engine in an SSH environment | |
582 | without a shared filesystem. You simply need to make sure that the controller is listening |
|
582 | without a shared filesystem. You simply need to make sure that the controller is listening | |
583 | on an interface visible to the engines, and move the connection file from the controller to |
|
583 | on an interface visible to the engines, and move the connection file from the controller to | |
584 | the engines. |
|
584 | the engines. | |
585 |
|
585 | |||
586 | 1. start the controller, listening on an ip-address visible to the engine machines:: |
|
586 | 1. start the controller, listening on an ip-address visible to the engine machines:: | |
587 |
|
587 | |||
588 | [controller.host] $ ipcontroller --ip=192.168.1.16 |
|
588 | [controller.host] $ ipcontroller --ip=192.168.1.16 | |
589 |
|
589 | |||
590 | [IPControllerApp] Using existing profile dir: u'/Users/me/.ipython/profile_default' |
|
590 | [IPControllerApp] Using existing profile dir: u'/Users/me/.ipython/profile_default' | |
591 | [IPControllerApp] Hub listening on tcp://192.168.1.16:63320 for registration. |
|
591 | [IPControllerApp] Hub listening on tcp://192.168.1.16:63320 for registration. | |
592 | [IPControllerApp] Hub using DB backend: 'IPython.parallel.controller.dictdb.DictDB' |
|
592 | [IPControllerApp] Hub using DB backend: 'IPython.parallel.controller.dictdb.DictDB' | |
593 | [IPControllerApp] hub::created hub |
|
593 | [IPControllerApp] hub::created hub | |
594 | [IPControllerApp] writing connection info to /Users/me/.ipython/profile_default/security/ipcontroller-client.json |
|
594 | [IPControllerApp] writing connection info to /Users/me/.ipython/profile_default/security/ipcontroller-client.json | |
595 | [IPControllerApp] writing connection info to /Users/me/.ipython/profile_default/security/ipcontroller-engine.json |
|
595 | [IPControllerApp] writing connection info to /Users/me/.ipython/profile_default/security/ipcontroller-engine.json | |
596 | [IPControllerApp] task::using Python leastload Task scheduler |
|
596 | [IPControllerApp] task::using Python leastload Task scheduler | |
597 | [IPControllerApp] Heartmonitor started |
|
597 | [IPControllerApp] Heartmonitor started | |
598 | [IPControllerApp] Creating pid file: /Users/me/.ipython/profile_default/pid/ipcontroller.pid |
|
598 | [IPControllerApp] Creating pid file: /Users/me/.ipython/profile_default/pid/ipcontroller.pid | |
599 | Scheduler started [leastload] |
|
599 | Scheduler started [leastload] | |
600 |
|
600 | |||
601 | 2. on each engine, fetch the connection file with scp:: |
|
601 | 2. on each engine, fetch the connection file with scp:: | |
602 |
|
602 | |||
603 | [engine.host.n] $ scp controller.host:.ipython/profile_default/security/ipcontroller-engine.json ./ |
|
603 | [engine.host.n] $ scp controller.host:.ipython/profile_default/security/ipcontroller-engine.json ./ | |
604 |
|
604 | |||
605 | .. note:: |
|
605 | .. note:: | |
606 |
|
606 | |||
607 | The log output of ipcontroller above shows you where the json files were written. |
|
607 | The log output of ipcontroller above shows you where the json files were written. | |
608 | They will be in :file:`~/.ipython` (or :file:`~/.config/ipython`) under |
|
608 | They will be in :file:`~/.ipython` (or :file:`~/.config/ipython`) under | |
609 | :file:`profile_default/security/ipcontroller-engine.json` |
|
609 | :file:`profile_default/security/ipcontroller-engine.json` | |
610 |
|
610 | |||
611 | 3. start the engines, using the connection file:: |
|
611 | 3. start the engines, using the connection file:: | |
612 |
|
612 | |||
613 | [engine.host.n] $ ipengine --file=./ipcontroller-engine.json |
|
613 | [engine.host.n] $ ipengine --file=./ipcontroller-engine.json | |
614 |
|
614 | |||
615 | A couple of notes: |
|
615 | A couple of notes: | |
616 |
|
616 | |||
617 | * You can avoid having to fetch the connection file every time by adding ``--reuse`` flag |
|
617 | * You can avoid having to fetch the connection file every time by adding ``--reuse`` flag | |
618 | to ipcontroller, which instructs the controller to read the previous connection file for |
|
618 | to ipcontroller, which instructs the controller to read the previous connection file for | |
619 | connection info, rather than generate a new one with randomized ports. |
|
619 | connection info, rather than generate a new one with randomized ports. | |
620 |
|
620 | |||
621 | * In step 2, if you fetch the connection file directly into the security dir of a profile, |
|
621 | * In step 2, if you fetch the connection file directly into the security dir of a profile, | |
622 | then you need not specify its path directly, only the profile (assumes the path exists, |
|
622 | then you need not specify its path directly, only the profile (assumes the path exists, | |
623 | otherwise you must create it first):: |
|
623 | otherwise you must create it first):: | |
624 |
|
624 | |||
625 | [engine.host.n] $ scp controller.host:.ipython/profile_default/security/ipcontroller-engine.json ~/.ipython/profile_ssh/security/ |
|
625 | [engine.host.n] $ scp controller.host:.ipython/profile_default/security/ipcontroller-engine.json ~/.ipython/profile_ssh/security/ | |
626 | [engine.host.n] $ ipengine --profile=ssh |
|
626 | [engine.host.n] $ ipengine --profile=ssh | |
627 |
|
627 | |||
628 | Of course, if you fetch the file into the default profile, no arguments must be passed to |
|
628 | Of course, if you fetch the file into the default profile, no arguments must be passed to | |
629 | ipengine at all. |
|
629 | ipengine at all. | |
630 |
|
630 | |||
631 | * Note that ipengine *did not* specify the ip argument. In general, it is unlikely for any |
|
631 | * Note that ipengine *did not* specify the ip argument. In general, it is unlikely for any | |
632 | connection information to be specified at the command-line to ipengine, as all of this |
|
632 | connection information to be specified at the command-line to ipengine, as all of this | |
633 | information should be contained in the connection file written by ipcontroller. |
|
633 | information should be contained in the connection file written by ipcontroller. | |
634 |
|
634 | |||
635 | Make JSON files persistent |
|
635 | Make JSON files persistent | |
636 | -------------------------- |
|
636 | -------------------------- | |
637 |
|
637 | |||
638 | At fist glance it may seem that that managing the JSON files is a bit |
|
638 | At fist glance it may seem that that managing the JSON files is a bit | |
639 | annoying. Going back to the house and key analogy, copying the JSON around |
|
639 | annoying. Going back to the house and key analogy, copying the JSON around | |
640 | each time you start the controller is like having to make a new key every time |
|
640 | each time you start the controller is like having to make a new key every time | |
641 | you want to unlock the door and enter your house. As with your house, you want |
|
641 | you want to unlock the door and enter your house. As with your house, you want | |
642 | to be able to create the key (or JSON file) once, and then simply use it at |
|
642 | to be able to create the key (or JSON file) once, and then simply use it at | |
643 | any point in the future. |
|
643 | any point in the future. | |
644 |
|
644 | |||
645 | To do this, the only thing you have to do is specify the `--reuse` flag, so that |
|
645 | To do this, the only thing you have to do is specify the `--reuse` flag, so that | |
646 | the connection information in the JSON files remains accurate:: |
|
646 | the connection information in the JSON files remains accurate:: | |
647 |
|
647 | |||
648 | $ ipcontroller --reuse |
|
648 | $ ipcontroller --reuse | |
649 |
|
649 | |||
650 | Then, just copy the JSON files over the first time and you are set. You can |
|
650 | Then, just copy the JSON files over the first time and you are set. You can | |
651 | start and stop the controller and engines any many times as you want in the |
|
651 | start and stop the controller and engines any many times as you want in the | |
652 | future, just make sure to tell the controller to reuse the file. |
|
652 | future, just make sure to tell the controller to reuse the file. | |
653 |
|
653 | |||
654 | .. note:: |
|
654 | .. note:: | |
655 |
|
655 | |||
656 | You may ask the question: what ports does the controller listen on if you |
|
656 | You may ask the question: what ports does the controller listen on if you | |
657 | don't tell is to use specific ones? The default is to use high random port |
|
657 | don't tell is to use specific ones? The default is to use high random port | |
658 | numbers. We do this for two reasons: i) to increase security through |
|
658 | numbers. We do this for two reasons: i) to increase security through | |
659 | obscurity and ii) to multiple controllers on a given host to start and |
|
659 | obscurity and ii) to multiple controllers on a given host to start and | |
660 | automatically use different ports. |
|
660 | automatically use different ports. | |
661 |
|
661 | |||
662 | Log files |
|
662 | Log files | |
663 | --------- |
|
663 | --------- | |
664 |
|
664 | |||
665 | All of the components of IPython have log files associated with them. |
|
665 | All of the components of IPython have log files associated with them. | |
666 | These log files can be extremely useful in debugging problems with |
|
666 | These log files can be extremely useful in debugging problems with | |
667 | IPython and can be found in the directory :file:`~/.ipython/profile_<name>/log`. |
|
667 | IPython and can be found in the directory :file:`~/.ipython/profile_<name>/log`. | |
668 | Sending the log files to us will often help us to debug any problems. |
|
668 | Sending the log files to us will often help us to debug any problems. | |
669 |
|
669 | |||
670 |
|
670 | |||
671 | Configuring `ipcontroller` |
|
671 | Configuring `ipcontroller` | |
672 | --------------------------- |
|
672 | --------------------------- | |
673 |
|
673 | |||
674 | The IPython Controller takes its configuration from the file :file:`ipcontroller_config.py` |
|
674 | The IPython Controller takes its configuration from the file :file:`ipcontroller_config.py` | |
675 | in the active profile directory. |
|
675 | in the active profile directory. | |
676 |
|
676 | |||
677 | Ports and addresses |
|
677 | Ports and addresses | |
678 | ******************* |
|
678 | ******************* | |
679 |
|
679 | |||
680 | In many cases, you will want to configure the Controller's network identity. By default, |
|
680 | In many cases, you will want to configure the Controller's network identity. By default, | |
681 | the Controller listens only on loopback, which is the most secure but often impractical. |
|
681 | the Controller listens only on loopback, which is the most secure but often impractical. | |
682 | To instruct the controller to listen on a specific interface, you can set the |
|
682 | To instruct the controller to listen on a specific interface, you can set the | |
683 | :attr:`HubFactory.ip` trait. To listen on all interfaces, simply specify: |
|
683 | :attr:`HubFactory.ip` trait. To listen on all interfaces, simply specify: | |
684 |
|
684 | |||
685 | .. sourcecode:: python |
|
685 | .. sourcecode:: python | |
686 |
|
686 | |||
687 | c.HubFactory.ip = '*' |
|
687 | c.HubFactory.ip = '*' | |
688 |
|
688 | |||
689 | When connecting to a Controller that is listening on loopback or behind a firewall, it may |
|
689 | When connecting to a Controller that is listening on loopback or behind a firewall, it may | |
690 | be necessary to specify an SSH server to use for tunnels, and the external IP of the |
|
690 | be necessary to specify an SSH server to use for tunnels, and the external IP of the | |
691 | Controller. If you specified that the HubFactory listen on loopback, or all interfaces, |
|
691 | Controller. If you specified that the HubFactory listen on loopback, or all interfaces, | |
692 | then IPython will try to guess the external IP. If you are on a system with VM network |
|
692 | then IPython will try to guess the external IP. If you are on a system with VM network | |
693 | devices, or many interfaces, this guess may be incorrect. In these cases, you will want |
|
693 | devices, or many interfaces, this guess may be incorrect. In these cases, you will want | |
694 | to specify the 'location' of the Controller. This is the IP of the machine the Controller |
|
694 | to specify the 'location' of the Controller. This is the IP of the machine the Controller | |
695 | is on, as seen by the clients, engines, or the SSH server used to tunnel connections. |
|
695 | is on, as seen by the clients, engines, or the SSH server used to tunnel connections. | |
696 |
|
696 | |||
697 | For example, to set up a cluster with a Controller on a work node, using ssh tunnels |
|
697 | For example, to set up a cluster with a Controller on a work node, using ssh tunnels | |
698 | through the login node, an example :file:`ipcontroller_config.py` might contain: |
|
698 | through the login node, an example :file:`ipcontroller_config.py` might contain: | |
699 |
|
699 | |||
700 | .. sourcecode:: python |
|
700 | .. sourcecode:: python | |
701 |
|
701 | |||
702 | # allow connections on all interfaces from engines |
|
702 | # allow connections on all interfaces from engines | |
703 | # engines on the same node will use loopback, while engines |
|
703 | # engines on the same node will use loopback, while engines | |
704 | # from other nodes will use an external IP |
|
704 | # from other nodes will use an external IP | |
705 | c.HubFactory.ip = '*' |
|
705 | c.HubFactory.ip = '*' | |
706 |
|
706 | |||
707 | # you typically only need to specify the location when there are extra |
|
707 | # you typically only need to specify the location when there are extra | |
708 | # interfaces that may not be visible to peer nodes (e.g. VM interfaces) |
|
708 | # interfaces that may not be visible to peer nodes (e.g. VM interfaces) | |
709 | c.HubFactory.location = '10.0.1.5' |
|
709 | c.HubFactory.location = '10.0.1.5' | |
710 | # or to get an automatic value, try this: |
|
710 | # or to get an automatic value, try this: | |
711 | import socket |
|
711 | import socket | |
712 | ex_ip = socket.gethostbyname_ex(socket.gethostname())[-1][0] |
|
712 | ex_ip = socket.gethostbyname_ex(socket.gethostname())[-1][0] | |
713 | c.HubFactory.location = ex_ip |
|
713 | c.HubFactory.location = ex_ip | |
714 |
|
714 | |||
715 | # now instruct clients to use the login node for SSH tunnels: |
|
715 | # now instruct clients to use the login node for SSH tunnels: | |
716 | c.HubFactory.ssh_server = 'login.mycluster.net' |
|
716 | c.HubFactory.ssh_server = 'login.mycluster.net' | |
717 |
|
717 | |||
718 | After doing this, your :file:`ipcontroller-client.json` file will look something like this: |
|
718 | After doing this, your :file:`ipcontroller-client.json` file will look something like this: | |
719 |
|
719 | |||
720 | .. this can be Python, despite the fact that it's actually JSON, because it's |
|
720 | .. this can be Python, despite the fact that it's actually JSON, because it's | |
721 | .. still valid Python |
|
721 | .. still valid Python | |
722 |
|
722 | |||
723 | .. sourcecode:: python |
|
723 | .. sourcecode:: python | |
724 |
|
724 | |||
725 | { |
|
725 | { | |
726 | "url":"tcp:\/\/*:43447", |
|
726 | "url":"tcp:\/\/*:43447", | |
727 | "exec_key":"9c7779e4-d08a-4c3b-ba8e-db1f80b562c1", |
|
727 | "exec_key":"9c7779e4-d08a-4c3b-ba8e-db1f80b562c1", | |
728 | "ssh":"login.mycluster.net", |
|
728 | "ssh":"login.mycluster.net", | |
729 | "location":"10.0.1.5" |
|
729 | "location":"10.0.1.5" | |
730 | } |
|
730 | } | |
731 |
|
731 | |||
732 | Then this file will be all you need for a client to connect to the controller, tunneling |
|
732 | Then this file will be all you need for a client to connect to the controller, tunneling | |
733 | SSH connections through login.mycluster.net. |
|
733 | SSH connections through login.mycluster.net. | |
734 |
|
734 | |||
735 | Database Backend |
|
735 | Database Backend | |
736 | **************** |
|
736 | **************** | |
737 |
|
737 | |||
738 | The Hub stores all messages and results passed between Clients and Engines. |
|
738 | The Hub stores all messages and results passed between Clients and Engines. | |
739 | For large and/or long-running clusters, it would be unreasonable to keep all |
|
739 | For large and/or long-running clusters, it would be unreasonable to keep all | |
740 | of this information in memory. For this reason, we have two database backends: |
|
740 | of this information in memory. For this reason, we have two database backends: | |
741 | [MongoDB]_ via PyMongo_, and SQLite with the stdlib :py:mod:`sqlite`. |
|
741 | [MongoDB]_ via PyMongo_, and SQLite with the stdlib :py:mod:`sqlite`. | |
742 |
|
742 | |||
743 | MongoDB is our design target, and the dict-like model it uses has driven our design. As far |
|
743 | MongoDB is our design target, and the dict-like model it uses has driven our design. As far | |
744 | as we are concerned, BSON can be considered essentially the same as JSON, adding support |
|
744 | as we are concerned, BSON can be considered essentially the same as JSON, adding support | |
745 | for binary data and datetime objects, and any new database backend must support the same |
|
745 | for binary data and datetime objects, and any new database backend must support the same | |
746 | data types. |
|
746 | data types. | |
747 |
|
747 | |||
748 | .. seealso:: |
|
748 | .. seealso:: | |
749 |
|
749 | |||
750 | MongoDB `BSON doc <http://www.mongodb.org/display/DOCS/BSON>`_ |
|
750 | MongoDB `BSON doc <http://www.mongodb.org/display/DOCS/BSON>`_ | |
751 |
|
751 | |||
752 | To use one of these backends, you must set the :attr:`HubFactory.db_class` trait: |
|
752 | To use one of these backends, you must set the :attr:`HubFactory.db_class` trait: | |
753 |
|
753 | |||
754 | .. sourcecode:: python |
|
754 | .. sourcecode:: python | |
755 |
|
755 | |||
756 | # for a simple dict-based in-memory implementation, use dictdb |
|
756 | # for a simple dict-based in-memory implementation, use dictdb | |
757 | # This is the default and the fastest, since it doesn't involve the filesystem |
|
757 | # This is the default and the fastest, since it doesn't involve the filesystem | |
758 | c.HubFactory.db_class = 'IPython.parallel.controller.dictdb.DictDB' |
|
758 | c.HubFactory.db_class = 'IPython.parallel.controller.dictdb.DictDB' | |
759 |
|
759 | |||
760 | # To use MongoDB: |
|
760 | # To use MongoDB: | |
761 | c.HubFactory.db_class = 'IPython.parallel.controller.mongodb.MongoDB' |
|
761 | c.HubFactory.db_class = 'IPython.parallel.controller.mongodb.MongoDB' | |
762 |
|
762 | |||
763 | # and SQLite: |
|
763 | # and SQLite: | |
764 | c.HubFactory.db_class = 'IPython.parallel.controller.sqlitedb.SQLiteDB' |
|
764 | c.HubFactory.db_class = 'IPython.parallel.controller.sqlitedb.SQLiteDB' | |
765 |
|
765 | |||
|
766 | # You can use NoDB to disable the database altogether, in case you don't need | |||
|
767 | # to reuse tasks or results, and want to keep memory consumption under control. | |||
|
768 | c.HubFactory.db_class = 'IPython.parallel.controller.dictdb.NoDB' | |||
|
769 | ||||
766 | When using the proper databases, you can actually allow for tasks to persist from |
|
770 | When using the proper databases, you can actually allow for tasks to persist from | |
767 | one session to the next by specifying the MongoDB database or SQLite table in |
|
771 | one session to the next by specifying the MongoDB database or SQLite table in | |
768 | which tasks are to be stored. The default is to use a table named for the Hub's Session, |
|
772 | which tasks are to be stored. The default is to use a table named for the Hub's Session, | |
769 | which is a UUID, and thus different every time. |
|
773 | which is a UUID, and thus different every time. | |
770 |
|
774 | |||
771 | .. sourcecode:: python |
|
775 | .. sourcecode:: python | |
772 |
|
776 | |||
773 | # To keep persistant task history in MongoDB: |
|
777 | # To keep persistant task history in MongoDB: | |
774 | c.MongoDB.database = 'tasks' |
|
778 | c.MongoDB.database = 'tasks' | |
775 |
|
779 | |||
776 | # and in SQLite: |
|
780 | # and in SQLite: | |
777 | c.SQLiteDB.table = 'tasks' |
|
781 | c.SQLiteDB.table = 'tasks' | |
778 |
|
782 | |||
779 |
|
783 | |||
780 | Since MongoDB servers can be running remotely or configured to listen on a particular port, |
|
784 | Since MongoDB servers can be running remotely or configured to listen on a particular port, | |
781 | you can specify any arguments you may need to the PyMongo `Connection |
|
785 | you can specify any arguments you may need to the PyMongo `Connection | |
782 | <http://api.mongodb.org/python/1.9/api/pymongo/connection.html#pymongo.connection.Connection>`_: |
|
786 | <http://api.mongodb.org/python/1.9/api/pymongo/connection.html#pymongo.connection.Connection>`_: | |
783 |
|
787 | |||
784 | .. sourcecode:: python |
|
788 | .. sourcecode:: python | |
785 |
|
789 | |||
786 | # positional args to pymongo.Connection |
|
790 | # positional args to pymongo.Connection | |
787 | c.MongoDB.connection_args = [] |
|
791 | c.MongoDB.connection_args = [] | |
788 |
|
792 | |||
789 | # keyword args to pymongo.Connection |
|
793 | # keyword args to pymongo.Connection | |
790 | c.MongoDB.connection_kwargs = {} |
|
794 | c.MongoDB.connection_kwargs = {} | |
791 |
|
795 | |||
|
796 | But sometimes you are moving lots of data around quickly, and you don't need | |||
|
797 | that information to be stored for later access, even by other Clients to this | |||
|
798 | same session. For this case, we have a dummy database, which doesn't actually | |||
|
799 | store anything. This lets the Hub stay small in memory, at the obvious expense | |||
|
800 | of being able to access the information that would have been stored in the | |||
|
801 | database (used for task resubmission, requesting results of tasks you didn't | |||
|
802 | submit, etc.). To use this backend, simply pass ``--nodb`` to | |||
|
803 | :command:`ipcontroller` on the command-line, or specify the :class:`NoDB` class | |||
|
804 | in your :file:`ipcontroller_config.py` as described above. | |||
|
805 | ||||
|
806 | ||||
|
807 | .. seealso:: | |||
|
808 | ||||
|
809 | For more information on the database backends, see the :ref:`db backend reference <parallel_db>`. | |||
|
810 | ||||
|
811 | ||||
792 | .. _PyMongo: http://api.mongodb.org/python/1.9/ |
|
812 | .. _PyMongo: http://api.mongodb.org/python/1.9/ | |
793 |
|
813 | |||
794 | Configuring `ipengine` |
|
814 | Configuring `ipengine` | |
795 | ----------------------- |
|
815 | ----------------------- | |
796 |
|
816 | |||
797 | The IPython Engine takes its configuration from the file :file:`ipengine_config.py` |
|
817 | The IPython Engine takes its configuration from the file :file:`ipengine_config.py` | |
798 |
|
818 | |||
799 | The Engine itself also has some amount of configuration. Most of this |
|
819 | The Engine itself also has some amount of configuration. Most of this | |
800 | has to do with initializing MPI or connecting to the controller. |
|
820 | has to do with initializing MPI or connecting to the controller. | |
801 |
|
821 | |||
802 | To instruct the Engine to initialize with an MPI environment set up by |
|
822 | To instruct the Engine to initialize with an MPI environment set up by | |
803 | mpi4py, add: |
|
823 | mpi4py, add: | |
804 |
|
824 | |||
805 | .. sourcecode:: python |
|
825 | .. sourcecode:: python | |
806 |
|
826 | |||
807 | c.MPI.use = 'mpi4py' |
|
827 | c.MPI.use = 'mpi4py' | |
808 |
|
828 | |||
809 | In this case, the Engine will use our default mpi4py init script to set up |
|
829 | In this case, the Engine will use our default mpi4py init script to set up | |
810 | the MPI environment prior to exection. We have default init scripts for |
|
830 | the MPI environment prior to exection. We have default init scripts for | |
811 | mpi4py and pytrilinos. If you want to specify your own code to be run |
|
831 | mpi4py and pytrilinos. If you want to specify your own code to be run | |
812 | at the beginning, specify `c.MPI.init_script`. |
|
832 | at the beginning, specify `c.MPI.init_script`. | |
813 |
|
833 | |||
814 | You can also specify a file or python command to be run at startup of the |
|
834 | You can also specify a file or python command to be run at startup of the | |
815 | Engine: |
|
835 | Engine: | |
816 |
|
836 | |||
817 | .. sourcecode:: python |
|
837 | .. sourcecode:: python | |
818 |
|
838 | |||
819 | c.IPEngineApp.startup_script = u'/path/to/my/startup.py' |
|
839 | c.IPEngineApp.startup_script = u'/path/to/my/startup.py' | |
820 |
|
840 | |||
821 | c.IPEngineApp.startup_command = 'import numpy, scipy, mpi4py' |
|
841 | c.IPEngineApp.startup_command = 'import numpy, scipy, mpi4py' | |
822 |
|
842 | |||
823 | These commands/files will be run again, after each |
|
843 | These commands/files will be run again, after each | |
824 |
|
844 | |||
825 | It's also useful on systems with shared filesystems to run the engines |
|
845 | It's also useful on systems with shared filesystems to run the engines | |
826 | in some scratch directory. This can be set with: |
|
846 | in some scratch directory. This can be set with: | |
827 |
|
847 | |||
828 | .. sourcecode:: python |
|
848 | .. sourcecode:: python | |
829 |
|
849 | |||
830 | c.IPEngineApp.work_dir = u'/path/to/scratch/' |
|
850 | c.IPEngineApp.work_dir = u'/path/to/scratch/' | |
831 |
|
851 | |||
832 |
|
852 | |||
833 |
|
853 | |||
834 | .. [MongoDB] MongoDB database http://www.mongodb.org |
|
854 | .. [MongoDB] MongoDB database http://www.mongodb.org | |
835 |
|
855 | |||
836 | .. [PBS] Portable Batch System http://www.openpbs.org |
|
856 | .. [PBS] Portable Batch System http://www.openpbs.org | |
837 |
|
857 | |||
838 | .. [SSH] SSH-Agent http://en.wikipedia.org/wiki/ssh-agent |
|
858 | .. [SSH] SSH-Agent http://en.wikipedia.org/wiki/ssh-agent |
General Comments 0
You need to be logged in to leave comments.
Login now