Show More
@@ -1,317 +1,317 b'' | |||
|
1 | 1 | """A Task logger that presents our DB interface, |
|
2 | 2 | but exists entirely in memory and implemented with dicts. |
|
3 | 3 | |
|
4 | 4 | Authors: |
|
5 | 5 | |
|
6 | 6 | * Min RK |
|
7 | 7 | |
|
8 | 8 | |
|
9 | 9 | TaskRecords are dicts of the form:: |
|
10 | 10 | |
|
11 | 11 | { |
|
12 | 12 | 'msg_id' : str(uuid), |
|
13 | 13 | 'client_uuid' : str(uuid), |
|
14 | 14 | 'engine_uuid' : str(uuid) or None, |
|
15 | 15 | 'header' : dict(header), |
|
16 | 16 | 'content': dict(content), |
|
17 | 17 | 'buffers': list(buffers), |
|
18 | 'submitted': datetime, | |
|
18 | 'submitted': datetime or None, | |
|
19 | 19 | 'started': datetime or None, |
|
20 | 20 | 'completed': datetime or None, |
|
21 |
're |
|
|
21 | 'received': datetime or None, | |
|
22 | 'resubmitted': str(uuid) or None, | |
|
22 | 23 | 'result_header' : dict(header) or None, |
|
23 | 24 | 'result_content' : dict(content) or None, |
|
24 | 25 | 'result_buffers' : list(buffers) or None, |
|
25 | 26 | } |
|
26 | 27 | |
|
27 | 28 | With this info, many of the special categories of tasks can be defined by query, |
|
28 | 29 | e.g.: |
|
29 | 30 | |
|
30 | 31 | * pending: completed is None |
|
31 | 32 | * client's outstanding: client_uuid = uuid && completed is None |
|
32 | 33 | * MIA: arrived is None (and completed is None) |
|
33 | 34 | |
|
34 | EngineRecords are dicts of the form:: | |
|
35 | ||
|
36 | { | |
|
37 | 'eid' : int(id), | |
|
38 | 'uuid': str(uuid) | |
|
39 | } | |
|
40 | ||
|
41 | This may be extended, but is currently. | |
|
42 | ||
|
43 | We support a subset of mongodb operators:: | |
|
35 | DictDB supports a subset of mongodb operators:: | |
|
44 | 36 | |
|
45 | 37 | $lt,$gt,$lte,$gte,$ne,$in,$nin,$all,$mod,$exists |
|
46 | 38 | """ |
|
47 | 39 | #----------------------------------------------------------------------------- |
|
48 | 40 | # Copyright (C) 2010-2011 The IPython Development Team |
|
49 | 41 | # |
|
50 | 42 | # Distributed under the terms of the BSD License. The full license is in |
|
51 | 43 | # the file COPYING, distributed as part of this software. |
|
52 | 44 | #----------------------------------------------------------------------------- |
|
53 | 45 | |
|
54 | 46 | from copy import deepcopy as copy |
|
55 | 47 | from datetime import datetime |
|
56 | 48 | |
|
57 | 49 | from IPython.config.configurable import LoggingConfigurable |
|
58 | 50 | |
|
59 | 51 | from IPython.utils.py3compat import iteritems, itervalues |
|
60 | 52 | from IPython.utils.traitlets import Dict, Unicode, Integer, Float |
|
61 | 53 | |
|
62 | 54 | filters = { |
|
63 | 55 | '$lt' : lambda a,b: a < b, |
|
64 | 56 | '$gt' : lambda a,b: b > a, |
|
65 | 57 | '$eq' : lambda a,b: a == b, |
|
66 | 58 | '$ne' : lambda a,b: a != b, |
|
67 | 59 | '$lte': lambda a,b: a <= b, |
|
68 | 60 | '$gte': lambda a,b: a >= b, |
|
69 | 61 | '$in' : lambda a,b: a in b, |
|
70 | 62 | '$nin': lambda a,b: a not in b, |
|
71 | 63 | '$all': lambda a,b: all([ a in bb for bb in b ]), |
|
72 | 64 | '$mod': lambda a,b: a%b[0] == b[1], |
|
73 | 65 | '$exists' : lambda a,b: (b and a is not None) or (a is None and not b) |
|
74 | 66 | } |
|
75 | 67 | |
|
76 | 68 | |
|
77 | 69 | class CompositeFilter(object): |
|
78 | 70 | """Composite filter for matching multiple properties.""" |
|
79 | 71 | |
|
80 | 72 | def __init__(self, dikt): |
|
81 | 73 | self.tests = [] |
|
82 | 74 | self.values = [] |
|
83 | 75 | for key, value in iteritems(dikt): |
|
84 | 76 | self.tests.append(filters[key]) |
|
85 | 77 | self.values.append(value) |
|
86 | 78 | |
|
87 | 79 | def __call__(self, value): |
|
88 | 80 | for test,check in zip(self.tests, self.values): |
|
89 | 81 | if not test(value, check): |
|
90 | 82 | return False |
|
91 | 83 | return True |
|
92 | 84 | |
|
93 | 85 | class BaseDB(LoggingConfigurable): |
|
94 | 86 | """Empty Parent class so traitlets work on DB.""" |
|
95 | 87 | # base configurable traits: |
|
96 | 88 | session = Unicode("") |
|
97 | 89 | |
|
98 | 90 | class DictDB(BaseDB): |
|
99 | 91 | """Basic in-memory dict-based object for saving Task Records. |
|
100 | 92 | |
|
101 | 93 | This is the first object to present the DB interface |
|
102 | 94 | for logging tasks out of memory. |
|
103 | 95 | |
|
104 | 96 | The interface is based on MongoDB, so adding a MongoDB |
|
105 | 97 | backend should be straightforward. |
|
106 | 98 | """ |
|
107 | 99 | |
|
108 | 100 | _records = Dict() |
|
109 | 101 | _culled_ids = set() # set of ids which have been culled |
|
110 | 102 | _buffer_bytes = Integer(0) # running total of the bytes in the DB |
|
111 | 103 | |
|
112 | 104 | size_limit = Integer(1024**3, config=True, |
|
113 | 105 | help="""The maximum total size (in bytes) of the buffers stored in the db |
|
114 | 106 | |
|
115 | 107 | When the db exceeds this size, the oldest records will be culled until |
|
116 | 108 | the total size is under size_limit * (1-cull_fraction). |
|
117 | 109 | default: 1 GB |
|
118 | 110 | """ |
|
119 | 111 | ) |
|
120 | 112 | record_limit = Integer(1024, config=True, |
|
121 | 113 | help="""The maximum number of records in the db |
|
122 | 114 | |
|
123 | 115 | When the history exceeds this size, the first record_limit * cull_fraction |
|
124 | 116 | records will be culled. |
|
125 | 117 | """ |
|
126 | 118 | ) |
|
127 | 119 | cull_fraction = Float(0.1, config=True, |
|
128 | 120 | help="""The fraction by which the db should culled when one of the limits is exceeded |
|
129 | 121 | |
|
130 | 122 | In general, the db size will spend most of its time with a size in the range: |
|
131 | 123 | |
|
132 | 124 | [limit * (1-cull_fraction), limit] |
|
133 | 125 | |
|
134 | 126 | for each of size_limit and record_limit. |
|
135 | 127 | """ |
|
136 | 128 | ) |
|
137 | 129 | |
|
138 | 130 | def _match_one(self, rec, tests): |
|
139 | 131 | """Check if a specific record matches tests.""" |
|
140 | 132 | for key,test in iteritems(tests): |
|
141 | 133 | if not test(rec.get(key, None)): |
|
142 | 134 | return False |
|
143 | 135 | return True |
|
144 | 136 | |
|
145 | 137 | def _match(self, check): |
|
146 | 138 | """Find all the matches for a check dict.""" |
|
147 | 139 | matches = [] |
|
148 | 140 | tests = {} |
|
149 | 141 | for k,v in iteritems(check): |
|
150 | 142 | if isinstance(v, dict): |
|
151 | 143 | tests[k] = CompositeFilter(v) |
|
152 | 144 | else: |
|
153 | 145 | tests[k] = lambda o: o==v |
|
154 | 146 | |
|
155 | 147 | for rec in itervalues(self._records): |
|
156 | 148 | if self._match_one(rec, tests): |
|
157 | 149 | matches.append(copy(rec)) |
|
158 | 150 | return matches |
|
159 | 151 | |
|
160 | 152 | def _extract_subdict(self, rec, keys): |
|
161 | 153 | """extract subdict of keys""" |
|
162 | 154 | d = {} |
|
163 | 155 | d['msg_id'] = rec['msg_id'] |
|
164 | 156 | for key in keys: |
|
165 | 157 | d[key] = rec[key] |
|
166 | 158 | return copy(d) |
|
167 | 159 | |
|
168 | 160 | # methods for monitoring size / culling history |
|
169 | 161 | |
|
170 | 162 | def _add_bytes(self, rec): |
|
171 | 163 | for key in ('buffers', 'result_buffers'): |
|
172 | 164 | for buf in rec.get(key) or []: |
|
173 | 165 | self._buffer_bytes += len(buf) |
|
174 | 166 | |
|
175 | 167 | self._maybe_cull() |
|
176 | 168 | |
|
177 | 169 | def _drop_bytes(self, rec): |
|
178 | 170 | for key in ('buffers', 'result_buffers'): |
|
179 | 171 | for buf in rec.get(key) or []: |
|
180 | 172 | self._buffer_bytes -= len(buf) |
|
181 | 173 | |
|
182 | 174 | def _cull_oldest(self, n=1): |
|
183 | 175 | """cull the oldest N records""" |
|
184 | 176 | for msg_id in self.get_history()[:n]: |
|
185 | 177 | self.log.debug("Culling record: %r", msg_id) |
|
186 | 178 | self._culled_ids.add(msg_id) |
|
187 | 179 | self.drop_record(msg_id) |
|
188 | 180 | |
|
189 | 181 | def _maybe_cull(self): |
|
190 | 182 | # cull by count: |
|
191 | 183 | if len(self._records) > self.record_limit: |
|
192 | 184 | to_cull = int(self.cull_fraction * self.record_limit) |
|
193 | 185 | self.log.info("%i records exceeds limit of %i, culling oldest %i", |
|
194 | 186 | len(self._records), self.record_limit, to_cull |
|
195 | 187 | ) |
|
196 | 188 | self._cull_oldest(to_cull) |
|
197 | 189 | |
|
198 | 190 | # cull by size: |
|
199 | 191 | if self._buffer_bytes > self.size_limit: |
|
200 | 192 | limit = self.size_limit * (1 - self.cull_fraction) |
|
201 | 193 | |
|
202 | 194 | before = self._buffer_bytes |
|
203 | 195 | before_count = len(self._records) |
|
204 | 196 | culled = 0 |
|
205 | 197 | while self._buffer_bytes > limit: |
|
206 | 198 | self._cull_oldest(1) |
|
207 | 199 | culled += 1 |
|
208 | 200 | |
|
209 | 201 | self.log.info("%i records with total buffer size %i exceeds limit: %i. Culled oldest %i records.", |
|
210 | 202 | before_count, before, self.size_limit, culled |
|
211 | 203 | ) |
|
212 | 204 | |
|
205 | def _check_dates(self, rec): | |
|
206 | for key in ('submitted', 'started', 'completed'): | |
|
207 | value = rec.get(key, None) | |
|
208 | if value is not None and not isinstance(value, datetime): | |
|
209 | raise ValueError("%s must be None or datetime, not %r" % (key, value)) | |
|
210 | ||
|
213 | 211 | # public API methods: |
|
214 | 212 | |
|
215 | 213 | def add_record(self, msg_id, rec): |
|
216 | 214 | """Add a new Task Record, by msg_id.""" |
|
217 | 215 | if msg_id in self._records: |
|
218 | 216 | raise KeyError("Already have msg_id %r"%(msg_id)) |
|
217 | self._check_dates(rec) | |
|
219 | 218 | self._records[msg_id] = rec |
|
220 | 219 | self._add_bytes(rec) |
|
221 | 220 | self._maybe_cull() |
|
222 | 221 | |
|
223 | 222 | def get_record(self, msg_id): |
|
224 | 223 | """Get a specific Task Record, by msg_id.""" |
|
225 | 224 | if msg_id in self._culled_ids: |
|
226 | 225 | raise KeyError("Record %r has been culled for size" % msg_id) |
|
227 | 226 | if not msg_id in self._records: |
|
228 | 227 | raise KeyError("No such msg_id %r"%(msg_id)) |
|
229 | 228 | return copy(self._records[msg_id]) |
|
230 | 229 | |
|
231 | 230 | def update_record(self, msg_id, rec): |
|
232 | 231 | """Update the data in an existing record.""" |
|
233 | 232 | if msg_id in self._culled_ids: |
|
234 | 233 | raise KeyError("Record %r has been culled for size" % msg_id) |
|
234 | self._check_dates(rec) | |
|
235 | 235 | _rec = self._records[msg_id] |
|
236 | 236 | self._drop_bytes(_rec) |
|
237 | 237 | _rec.update(rec) |
|
238 | 238 | self._add_bytes(_rec) |
|
239 | 239 | |
|
240 | 240 | def drop_matching_records(self, check): |
|
241 | 241 | """Remove a record from the DB.""" |
|
242 | 242 | matches = self._match(check) |
|
243 | 243 | for rec in matches: |
|
244 | 244 | self._drop_bytes(rec) |
|
245 | 245 | del self._records[rec['msg_id']] |
|
246 | 246 | |
|
247 | 247 | def drop_record(self, msg_id): |
|
248 | 248 | """Remove a record from the DB.""" |
|
249 | 249 | rec = self._records[msg_id] |
|
250 | 250 | self._drop_bytes(rec) |
|
251 | 251 | del self._records[msg_id] |
|
252 | 252 | |
|
253 | 253 | def find_records(self, check, keys=None): |
|
254 | 254 | """Find records matching a query dict, optionally extracting subset of keys. |
|
255 | 255 | |
|
256 | 256 | Returns dict keyed by msg_id of matching records. |
|
257 | 257 | |
|
258 | 258 | Parameters |
|
259 | 259 | ---------- |
|
260 | 260 | |
|
261 | 261 | check: dict |
|
262 | 262 | mongodb-style query argument |
|
263 | 263 | keys: list of strs [optional] |
|
264 | 264 | if specified, the subset of keys to extract. msg_id will *always* be |
|
265 | 265 | included. |
|
266 | 266 | """ |
|
267 | 267 | matches = self._match(check) |
|
268 | 268 | if keys: |
|
269 | 269 | return [ self._extract_subdict(rec, keys) for rec in matches ] |
|
270 | 270 | else: |
|
271 | 271 | return matches |
|
272 | 272 | |
|
273 | 273 | def get_history(self): |
|
274 | 274 | """get all msg_ids, ordered by time submitted.""" |
|
275 | 275 | msg_ids = self._records.keys() |
|
276 | 276 | # Remove any that do not have a submitted timestamp. |
|
277 | 277 | # This is extremely unlikely to happen, |
|
278 | 278 | # but it seems to come up in some tests on VMs. |
|
279 | 279 | msg_ids = [ m for m in msg_ids if self._records[m]['submitted'] is not None ] |
|
280 | 280 | return sorted(msg_ids, key=lambda m: self._records[m]['submitted']) |
|
281 | 281 | |
|
282 | 282 | |
|
283 | 283 | NODATA = KeyError("NoDB backend doesn't store any data. " |
|
284 | 284 | "Start the Controller with a DB backend to enable resubmission / result persistence." |
|
285 | 285 | ) |
|
286 | 286 | |
|
287 | 287 | |
|
288 | 288 | class NoDB(BaseDB): |
|
289 | 289 | """A blackhole db backend that actually stores no information. |
|
290 | 290 | |
|
291 | 291 | Provides the full DB interface, but raises KeyErrors on any |
|
292 | 292 | method that tries to access the records. This can be used to |
|
293 | 293 | minimize the memory footprint of the Hub when its record-keeping |
|
294 | 294 | functionality is not required. |
|
295 | 295 | """ |
|
296 | 296 | |
|
297 | 297 | def add_record(self, msg_id, record): |
|
298 | 298 | pass |
|
299 | 299 | |
|
300 | 300 | def get_record(self, msg_id): |
|
301 | 301 | raise NODATA |
|
302 | 302 | |
|
303 | 303 | def update_record(self, msg_id, record): |
|
304 | 304 | pass |
|
305 | 305 | |
|
306 | 306 | def drop_matching_records(self, check): |
|
307 | 307 | pass |
|
308 | 308 | |
|
309 | 309 | def drop_record(self, msg_id): |
|
310 | 310 | pass |
|
311 | 311 | |
|
312 | 312 | def find_records(self, check, keys=None): |
|
313 | 313 | raise NODATA |
|
314 | 314 | |
|
315 | 315 | def get_history(self): |
|
316 | 316 | raise NODATA |
|
317 | 317 |
General Comments 0
You need to be logged in to leave comments.
Login now