Show More
@@ -1,517 +1,520 b'' | |||
|
1 | 1 | """ |
|
2 | 2 | Gunicorn config extension and hooks. This config file adds some extra settings and memory management. |
|
3 | 3 | Gunicorn configuration should be managed by .ini files entries of RhodeCode or VCSServer |
|
4 | 4 | """ |
|
5 | 5 | |
|
6 | import contextlib | |
|
7 | import dataclasses | |
|
8 | 6 | import gc |
|
7 | import os | |
|
8 | import sys | |
|
9 | 9 | import math |
|
10 |
import |
|
|
10 | import time | |
|
11 | import threading | |
|
12 | import traceback | |
|
11 | 13 | import random |
|
12 | 14 | import socket |
|
13 |
import |
|
|
14 | import threading | |
|
15 | import time | |
|
16 | import traceback | |
|
17 | ||
|
15 | import dataclasses | |
|
18 | 16 | from gunicorn.glogging import Logger |
|
19 | 17 | |
|
20 | 18 | |
|
21 | 19 | def get_workers(): |
|
22 | 20 | import multiprocessing |
|
23 | ||
|
24 | 21 | return multiprocessing.cpu_count() * 2 + 1 |
|
25 | 22 | |
|
26 | 23 | |
|
27 | 24 | bind = "127.0.0.1:10010" |
|
28 | 25 | |
|
29 | 26 | |
|
30 | 27 | # Error logging output for gunicorn (-) is stdout |
|
31 |
errorlog = |
|
|
28 | errorlog = '-' | |
|
32 | 29 | |
|
33 | 30 | # Access logging output for gunicorn (-) is stdout |
|
34 |
accesslog = |
|
|
31 | accesslog = '-' | |
|
35 | 32 | |
|
36 | 33 | |
|
37 | 34 | # SERVER MECHANICS |
|
38 | 35 | # None == system temp dir |
|
39 | 36 | # worker_tmp_dir is recommended to be set to some tmpfs |
|
40 | 37 | worker_tmp_dir = None |
|
41 | 38 | tmp_upload_dir = None |
|
42 | 39 | |
|
43 | 40 | # use re-use port logic |
|
44 | 41 |
# |
|
45 | 42 | |
|
46 | 43 | # Custom log format |
|
47 | 44 |
# |
|
48 | 45 | # '%(t)s %(p)s INFO [GNCRN] %(h)-15s rqt:%(L)s %(s)s %(b)-6s "%(m)s:%(U)s %(q)s" usr:%(u)s "%(f)s" "%(a)s"') |
|
49 | 46 | |
|
50 | 47 | # loki format for easier parsing in grafana |
|
51 | access_log_format = 'time="%(t)s" pid=%(p)s level="INFO" type="[GNCRN]" ip="%(h)-15s" rqt="%(L)s" response_code="%(s)s" response_bytes="%(b)-6s" uri="%(m)s:%(U)s %(q)s" user=":%(u)s" user_agent="%(a)s"' | |
|
52 | ||
|
53 | ||
|
54 | # Sets the number of process workers. More workers means more concurrent connections | |
|
55 | # RhodeCode can handle at the same time. Each additional worker also it increases | |
|
56 | # memory usage as each has it's own set of caches. | |
|
57 | # Recommended value is (2 * NUMBER_OF_CPUS + 1), eg 2CPU = 5 workers, but no more | |
|
58 | # than 8-10 unless for huge deployments .e.g 700-1000 users. | |
|
59 | # `instance_id = *` must be set in the [app:main] section below (which is the default) | |
|
60 | # when using more than 1 worker. | |
|
61 | workers = 6 | |
|
48 | access_log_format = ( | |
|
49 | 'time="%(t)s" pid=%(p)s level="INFO" type="[GNCRN]" ip="%(h)-15s" rqt="%(L)s" response_code="%(s)s" response_bytes="%(b)-6s" uri="%(m)s:%(U)s %(q)s" user=":%(u)s" user_agent="%(a)s"') | |
|
62 | 50 | |
|
63 | 51 | # self adjust workers based on CPU count, to use maximum of CPU and not overquota the resources |
|
64 | 52 | # workers = get_workers() |
|
65 | 53 | |
|
66 | 54 | # Gunicorn access log level |
|
67 |
loglevel = |
|
|
55 | loglevel = 'info' | |
|
68 | 56 | |
|
69 | 57 | # Process name visible in a process list |
|
70 | 58 | proc_name = "rhodecode_vcsserver" |
|
71 | 59 | |
|
72 | # Type of worker class, one of `sync`, `gevent` | |
|
73 |
# currently `sync` is the only option allowed |
|
|
74 | worker_class = "sync" | |
|
60 | # Type of worker class, one of `sync`, `gevent` or `gthread` | |
|
61 | # currently `sync` is the only option allowed for vcsserver and for rhodecode all of 3 are allowed | |
|
62 | # gevent: | |
|
63 | # In this case, the maximum number of concurrent requests is (N workers * X worker_connections) | |
|
64 | # e.g. workers =3 worker_connections=10 = 3*10, 30 concurrent requests can be handled | |
|
65 | # gthread: | |
|
66 | # In this case, the maximum number of concurrent requests is (N workers * X threads) | |
|
67 | # e.g. workers = 3 threads=3 = 3*3, 9 concurrent requests can be handled | |
|
68 | worker_class = 'sync' | |
|
69 | ||
|
70 | # Sets the number of process workers. More workers means more concurrent connections | |
|
71 | # RhodeCode can handle at the same time. Each additional worker also it increases | |
|
72 | # memory usage as each has its own set of caches. | |
|
73 | # The Recommended value is (2 * NUMBER_OF_CPUS + 1), eg 2CPU = 5 workers, but no more | |
|
74 | # than 8-10 unless for huge deployments .e.g 700-1000 users. | |
|
75 | # `instance_id = *` must be set in the [app:main] section below (which is the default) | |
|
76 | # when using more than 1 worker. | |
|
77 | workers = 2 | |
|
78 | ||
|
79 | # Threads numbers for worker class gthread | |
|
80 | threads = 1 | |
|
75 | 81 | |
|
76 | 82 | # The maximum number of simultaneous clients. Valid only for gevent |
|
83 | # In this case, the maximum number of concurrent requests is (N workers * X worker_connections) | |
|
84 | # e.g workers =3 worker_connections=10 = 3*10, 30 concurrent requests can be handled | |
|
77 | 85 | worker_connections = 10 |
|
78 | 86 | |
|
79 | 87 | # Max number of requests that worker will handle before being gracefully restarted. |
|
80 | 88 | # Prevents memory leaks, jitter adds variability so not all workers are restarted at once. |
|
81 | 89 | max_requests = 2000 |
|
82 | 90 | max_requests_jitter = int(max_requests * 0.2) # 20% of max_requests |
|
83 | 91 | |
|
84 | 92 | # The maximum number of pending connections. |
|
85 | 93 | # Exceeding this number results in the client getting an error when attempting to connect. |
|
86 | 94 | backlog = 64 |
|
87 | 95 | |
|
88 | 96 | # The Amount of time a worker can spend with handling a request before it |
|
89 | 97 | # gets killed and restarted. By default, set to 21600 (6hrs) |
|
90 | 98 | # Examples: 1800 (30min), 3600 (1hr), 7200 (2hr), 43200 (12h) |
|
91 | 99 | timeout = 21600 |
|
92 | 100 | |
|
93 | 101 | # The maximum size of HTTP request line in bytes. |
|
94 | 102 | # 0 for unlimited |
|
95 | 103 | limit_request_line = 0 |
|
96 | 104 | |
|
97 | 105 | # Limit the number of HTTP headers fields in a request. |
|
98 | 106 | # By default this value is 100 and can't be larger than 32768. |
|
99 | 107 | limit_request_fields = 32768 |
|
100 | 108 | |
|
101 | 109 | # Limit the allowed size of an HTTP request header field. |
|
102 | 110 | # Value is a positive number or 0. |
|
103 | 111 | # Setting it to 0 will allow unlimited header field sizes. |
|
104 | 112 | limit_request_field_size = 0 |
|
105 | 113 | |
|
106 | 114 | # Timeout for graceful workers restart. |
|
107 | 115 | # After receiving a restart signal, workers have this much time to finish |
|
108 | 116 | # serving requests. Workers still alive after the timeout (starting from the |
|
109 | 117 | # receipt of the restart signal) are force killed. |
|
110 | 118 | # Examples: 1800 (30min), 3600 (1hr), 7200 (2hr), 43200 (12h) |
|
111 | 119 | graceful_timeout = 21600 |
|
112 | 120 | |
|
113 | 121 | # The number of seconds to wait for requests on a Keep-Alive connection. |
|
114 | 122 | # Generally set in the 1-5 seconds range. |
|
115 | 123 | keepalive = 2 |
|
116 | 124 | |
|
117 | 125 | # Maximum memory usage that each worker can use before it will receive a |
|
118 | 126 | # graceful restart signal 0 = memory monitoring is disabled |
|
119 | 127 | # Examples: 268435456 (256MB), 536870912 (512MB) |
|
120 | 128 | # 1073741824 (1GB), 2147483648 (2GB), 4294967296 (4GB) |
|
121 | 129 | # Dynamic formula 1024 * 1024 * 256 == 256MBs |
|
122 | 130 | memory_max_usage = 0 |
|
123 | 131 | |
|
124 | 132 | # How often in seconds to check for memory usage for each gunicorn worker |
|
125 | 133 | memory_usage_check_interval = 60 |
|
126 | 134 | |
|
127 | 135 | # Threshold value for which we don't recycle worker if GarbageCollection |
|
128 | 136 | # frees up enough resources. Before each restart, we try to run GC on worker |
|
129 | 137 | # in case we get enough free memory after that; restart will not happen. |
|
130 | 138 | memory_usage_recovery_threshold = 0.8 |
|
131 | 139 | |
|
132 | 140 | |
|
133 | 141 | @dataclasses.dataclass |
|
134 | 142 | class MemoryCheckConfig: |
|
135 | 143 | max_usage: int |
|
136 | 144 | check_interval: int |
|
137 | 145 | recovery_threshold: float |
|
138 | 146 | |
|
139 | 147 | |
|
140 | 148 | def _get_process_rss(pid=None): |
|
141 | with contextlib.suppress(Exception): | |
|
149 | try: | |
|
142 | 150 | import psutil |
|
143 | ||
|
144 | 151 | if pid: |
|
145 | 152 | proc = psutil.Process(pid) |
|
146 | 153 | else: |
|
147 | 154 | proc = psutil.Process() |
|
148 | 155 | return proc.memory_info().rss |
|
149 | ||
|
156 | except Exception: | |
|
150 | 157 | return None |
|
151 | 158 | |
|
152 | 159 | |
|
153 | 160 | def _get_config(ini_path): |
|
154 | 161 | import configparser |
|
155 | 162 | |
|
156 | with contextlib.suppress(Exception): | |
|
163 | try: | |
|
157 | 164 | config = configparser.RawConfigParser() |
|
158 | 165 | config.read(ini_path) |
|
159 | 166 | return config |
|
160 | ||
|
167 | except Exception: | |
|
161 | 168 | return None |
|
162 | 169 | |
|
163 | 170 | |
|
164 | 171 | def get_memory_usage_params(config=None): |
|
165 | 172 | # memory spec defaults |
|
166 | 173 | _memory_max_usage = memory_max_usage |
|
167 | 174 | _memory_usage_check_interval = memory_usage_check_interval |
|
168 | 175 | _memory_usage_recovery_threshold = memory_usage_recovery_threshold |
|
169 | 176 | |
|
170 | 177 | if config: |
|
171 | 178 | ini_path = os.path.abspath(config) |
|
172 | 179 | conf = _get_config(ini_path) |
|
173 | 180 | |
|
174 |
section = |
|
|
181 | section = 'server:main' | |
|
175 | 182 | if conf and conf.has_section(section): |
|
176 | if conf.has_option(section, "memory_max_usage"): | |
|
177 | _memory_max_usage = conf.getint(section, "memory_max_usage") | |
|
178 | 183 | |
|
179 |
if conf.has_option(section, |
|
|
180 |
_memory_usage |
|
|
184 | if conf.has_option(section, 'memory_max_usage'): | |
|
185 | _memory_max_usage = conf.getint(section, 'memory_max_usage') | |
|
186 | ||
|
187 | if conf.has_option(section, 'memory_usage_check_interval'): | |
|
188 | _memory_usage_check_interval = conf.getint(section, 'memory_usage_check_interval') | |
|
181 | 189 | |
|
182 |
if conf.has_option(section, |
|
|
183 |
_memory_usage_recovery_threshold = conf.getfloat(section, |
|
|
190 | if conf.has_option(section, 'memory_usage_recovery_threshold'): | |
|
191 | _memory_usage_recovery_threshold = conf.getfloat(section, 'memory_usage_recovery_threshold') | |
|
184 | 192 | |
|
185 |
_memory_max_usage = int(os.environ.get( |
|
|
186 | _memory_usage_check_interval = int( | |
|
187 |
|
|
|
188 | ) | |
|
189 | _memory_usage_recovery_threshold = float( | |
|
190 | os.environ.get("RC_GUNICORN_MEMORY_USAGE_RECOVERY_THRESHOLD", "") or _memory_usage_recovery_threshold | |
|
191 | ) | |
|
193 | _memory_max_usage = int(os.environ.get('RC_GUNICORN_MEMORY_MAX_USAGE', '') | |
|
194 | or _memory_max_usage) | |
|
195 | _memory_usage_check_interval = int(os.environ.get('RC_GUNICORN_MEMORY_USAGE_CHECK_INTERVAL', '') | |
|
196 | or _memory_usage_check_interval) | |
|
197 | _memory_usage_recovery_threshold = float(os.environ.get('RC_GUNICORN_MEMORY_USAGE_RECOVERY_THRESHOLD', '') | |
|
198 | or _memory_usage_recovery_threshold) | |
|
192 | 199 | |
|
193 | 200 | return MemoryCheckConfig(_memory_max_usage, _memory_usage_check_interval, _memory_usage_recovery_threshold) |
|
194 | 201 | |
|
195 | 202 | |
|
196 | 203 | def _time_with_offset(check_interval): |
|
197 | 204 |
return time.time() - random.randint(0, check_interval |
|
198 | 205 | |
|
199 | 206 | |
|
200 | 207 | def pre_fork(server, worker): |
|
201 | 208 | pass |
|
202 | 209 | |
|
203 | 210 | |
|
204 | 211 | def post_fork(server, worker): |
|
212 | ||
|
205 | 213 | memory_conf = get_memory_usage_params() |
|
206 | 214 | _memory_max_usage = memory_conf.max_usage |
|
207 | 215 | _memory_usage_check_interval = memory_conf.check_interval |
|
208 | 216 | _memory_usage_recovery_threshold = memory_conf.recovery_threshold |
|
209 | 217 | |
|
210 |
worker._memory_max_usage = int(os.environ.get( |
|
|
211 | worker._memory_usage_check_interval = int( | |
|
212 |
|
|
|
213 | ) | |
|
214 | worker._memory_usage_recovery_threshold = float( | |
|
215 | os.environ.get("RC_GUNICORN_MEMORY_USAGE_RECOVERY_THRESHOLD", "") or _memory_usage_recovery_threshold | |
|
216 | ) | |
|
218 | worker._memory_max_usage = int(os.environ.get('RC_GUNICORN_MEMORY_MAX_USAGE', '') | |
|
219 | or _memory_max_usage) | |
|
220 | worker._memory_usage_check_interval = int(os.environ.get('RC_GUNICORN_MEMORY_USAGE_CHECK_INTERVAL', '') | |
|
221 | or _memory_usage_check_interval) | |
|
222 | worker._memory_usage_recovery_threshold = float(os.environ.get('RC_GUNICORN_MEMORY_USAGE_RECOVERY_THRESHOLD', '') | |
|
223 | or _memory_usage_recovery_threshold) | |
|
217 | 224 | |
|
218 | 225 | # register memory last check time, with some random offset so we don't recycle all |
|
219 | 226 | # at once |
|
220 | 227 | worker._last_memory_check_time = _time_with_offset(_memory_usage_check_interval) |
|
221 | 228 | |
|
222 | 229 | if _memory_max_usage: |
|
223 | server.log.info( | |
|
224 |
|
|
|
225 | ) | |
|
230 | server.log.info("pid=[%-10s] WORKER spawned with max memory set at %s", worker.pid, | |
|
231 | _format_data_size(_memory_max_usage)) | |
|
226 | 232 | else: |
|
227 | 233 | server.log.info("pid=[%-10s] WORKER spawned", worker.pid) |
|
228 | 234 | |
|
229 | 235 | |
|
230 | 236 | def pre_exec(server): |
|
231 | 237 | server.log.info("Forked child, re-executing.") |
|
232 | 238 | |
|
233 | 239 | |
|
234 | 240 | def on_starting(server): |
|
235 |
server_lbl = |
|
|
241 | server_lbl = '{} {}'.format(server.proc_name, server.address) | |
|
236 | 242 | server.log.info("Server %s is starting.", server_lbl) |
|
237 |
server.log.info( |
|
|
243 | server.log.info('Config:') | |
|
238 | 244 | server.log.info(f"\n{server.cfg}") |
|
239 | 245 | server.log.info(get_memory_usage_params()) |
|
240 | 246 | |
|
241 | 247 | |
|
242 | 248 | def when_ready(server): |
|
243 | 249 | server.log.info("Server %s is ready. Spawning workers", server) |
|
244 | 250 | |
|
245 | 251 | |
|
246 | 252 | def on_reload(server): |
|
247 | 253 | pass |
|
248 | 254 | |
|
249 | 255 | |
|
250 | 256 | def _format_data_size(size, unit="B", precision=1, binary=True): |
|
251 | 257 | """Format a number using SI units (kilo, mega, etc.). |
|
252 | 258 | |
|
253 | 259 | ``size``: The number as a float or int. |
|
254 | 260 | |
|
255 | 261 | ``unit``: The unit name in plural form. Examples: "bytes", "B". |
|
256 | 262 | |
|
257 | 263 | ``precision``: How many digits to the right of the decimal point. Default |
|
258 | 264 | is 1. 0 suppresses the decimal point. |
|
259 | 265 | |
|
260 | 266 | ``binary``: If false, use base-10 decimal prefixes (kilo = K = 1000). |
|
261 | 267 | If true, use base-2 binary prefixes (kibi = Ki = 1024). |
|
262 | 268 | |
|
263 | 269 | ``full_name``: If false (default), use the prefix abbreviation ("k" or |
|
264 | 270 | "Ki"). If true, use the full prefix ("kilo" or "kibi"). If false, |
|
265 | 271 | use abbreviation ("k" or "Ki"). |
|
266 | 272 | |
|
267 | 273 | """ |
|
268 | 274 | |
|
269 | 275 | if not binary: |
|
270 | 276 | base = 1000 |
|
271 |
multiples = ( |
|
|
277 | multiples = ('', 'k', 'M', 'G', 'T', 'P', 'E', 'Z', 'Y') | |
|
272 | 278 | else: |
|
273 | 279 | base = 1024 |
|
274 |
multiples = ( |
|
|
280 | multiples = ('', 'Ki', 'Mi', 'Gi', 'Ti', 'Pi', 'Ei', 'Zi', 'Yi') | |
|
275 | 281 | |
|
276 | 282 | sign = "" |
|
277 | 283 | if size > 0: |
|
278 | 284 | m = int(math.log(size, base)) |
|
279 | 285 | elif size < 0: |
|
280 | 286 | sign = "-" |
|
281 | 287 | size = -size |
|
282 | 288 | m = int(math.log(size, base)) |
|
283 | 289 | else: |
|
284 | 290 | m = 0 |
|
285 | 291 | if m > 8: |
|
286 | 292 | m = 8 |
|
287 | 293 | |
|
288 | 294 | if m == 0: |
|
289 |
precision = |
|
|
295 | precision = '%.0f' | |
|
290 | 296 | else: |
|
291 |
precision = |
|
|
297 | precision = '%%.%df' % precision | |
|
292 | 298 | |
|
293 | 299 | size = precision % (size / math.pow(base, m)) |
|
294 | 300 | |
|
295 |
return |
|
|
301 | return '%s%s %s%s' % (sign, size.strip(), multiples[m], unit) | |
|
296 | 302 | |
|
297 | 303 | |
|
298 | 304 | def _check_memory_usage(worker): |
|
299 | 305 | _memory_max_usage = worker._memory_max_usage |
|
300 | 306 | if not _memory_max_usage: |
|
301 | 307 | return |
|
302 | 308 | |
|
303 | 309 | _memory_usage_check_interval = worker._memory_usage_check_interval |
|
304 | 310 | _memory_usage_recovery_threshold = memory_max_usage * worker._memory_usage_recovery_threshold |
|
305 | 311 | |
|
306 | 312 | elapsed = time.time() - worker._last_memory_check_time |
|
307 | 313 | if elapsed > _memory_usage_check_interval: |
|
308 | 314 | mem_usage = _get_process_rss() |
|
309 | 315 | if mem_usage and mem_usage > _memory_max_usage: |
|
310 | 316 | worker.log.info( |
|
311 |
"memory usage %s > %s, forcing gc", |
|
|
312 | ) | |
|
317 | "memory usage %s > %s, forcing gc", | |
|
318 | _format_data_size(mem_usage), _format_data_size(_memory_max_usage)) | |
|
313 | 319 | # Try to clean it up by forcing a full collection. |
|
314 | 320 | gc.collect() |
|
315 | 321 | mem_usage = _get_process_rss() |
|
316 | 322 | if mem_usage > _memory_usage_recovery_threshold: |
|
317 | 323 | # Didn't clean up enough, we'll have to terminate. |
|
318 | 324 | worker.log.warning( |
|
319 | 325 | "memory usage %s > %s after gc, quitting", |
|
320 | _format_data_size(mem_usage), | |
|
321 | _format_data_size(_memory_max_usage), | |
|
322 | ) | |
|
326 | _format_data_size(mem_usage), _format_data_size(_memory_max_usage)) | |
|
323 | 327 | # This will cause worker to auto-restart itself |
|
324 | 328 | worker.alive = False |
|
325 | 329 | worker._last_memory_check_time = time.time() |
|
326 | 330 | |
|
327 | 331 | |
|
328 | 332 | def worker_int(worker): |
|
329 | 333 | worker.log.info("pid=[%-10s] worker received INT or QUIT signal", worker.pid) |
|
330 | 334 | |
|
331 | 335 | # get traceback info, when a worker crashes |
|
332 | 336 | def get_thread_id(t_id): |
|
333 | 337 | id2name = dict([(th.ident, th.name) for th in threading.enumerate()]) |
|
334 | 338 | return id2name.get(t_id, "unknown_thread_id") |
|
335 | 339 | |
|
336 | 340 | code = [] |
|
337 | 341 | for thread_id, stack in sys._current_frames().items(): # noqa |
|
338 | code.append("\n# Thread: %s(%d)" % (get_thread_id(thread_id), thread_id)) | |
|
342 | code.append( | |
|
343 | "\n# Thread: %s(%d)" % (get_thread_id(thread_id), thread_id)) | |
|
339 | 344 | for fname, lineno, name, line in traceback.extract_stack(stack): |
|
340 | 345 | code.append('File: "%s", line %d, in %s' % (fname, lineno, name)) |
|
341 | 346 | if line: |
|
342 | 347 | code.append(" %s" % (line.strip())) |
|
343 | 348 | worker.log.debug("\n".join(code)) |
|
344 | 349 | |
|
345 | 350 | |
|
346 | 351 | def worker_abort(worker): |
|
347 | 352 | worker.log.info("pid=[%-10s] worker received SIGABRT signal", worker.pid) |
|
348 | 353 | |
|
349 | 354 | |
|
350 | 355 | def worker_exit(server, worker): |
|
351 | 356 | worker.log.info("pid=[%-10s] worker exit", worker.pid) |
|
352 | 357 | |
|
353 | 358 | |
|
354 | 359 | def child_exit(server, worker): |
|
355 | 360 | worker.log.info("pid=[%-10s] worker child exit", worker.pid) |
|
356 | 361 | |
|
357 | 362 | |
|
358 | 363 | def pre_request(worker, req): |
|
359 | 364 | worker.start_time = time.time() |
|
360 | worker.log.debug("GNCRN PRE WORKER [cnt:%s]: %s %s", worker.nr, req.method, req.path) | |
|
365 | worker.log.debug( | |
|
366 | "GNCRN PRE WORKER [cnt:%s]: %s %s", worker.nr, req.method, req.path) | |
|
361 | 367 | |
|
362 | 368 | |
|
363 | 369 | def post_request(worker, req, environ, resp): |
|
364 | 370 | total_time = time.time() - worker.start_time |
|
365 | 371 | # Gunicorn sometimes has problems with reading the status_code |
|
366 |
status_code = getattr(resp, |
|
|
372 | status_code = getattr(resp, 'status_code', '') | |
|
367 | 373 | worker.log.debug( |
|
368 | 374 | "GNCRN POST WORKER [cnt:%s]: %s %s resp: %s, Load Time: %.4fs", |
|
369 | worker.nr, | |
|
370 | req.method, | |
|
371 | req.path, | |
|
372 | status_code, | |
|
373 | total_time, | |
|
374 | ) | |
|
375 | worker.nr, req.method, req.path, status_code, total_time) | |
|
375 | 376 | _check_memory_usage(worker) |
|
376 | 377 | |
|
377 | 378 | |
|
378 | 379 | def _filter_proxy(ip): |
|
379 | 380 | """ |
|
380 | 381 | Passed in IP addresses in HEADERS can be in a special format of multiple |
|
381 | 382 | ips. Those comma separated IPs are passed from various proxies in the |
|
382 | 383 | chain of request processing. The left-most being the original client. |
|
383 | 384 | We only care about the first IP which came from the org. client. |
|
384 | 385 | |
|
385 | 386 | :param ip: ip string from headers |
|
386 | 387 | """ |
|
387 |
if |
|
|
388 |
_ips = ip.split( |
|
|
388 | if ',' in ip: | |
|
389 | _ips = ip.split(',') | |
|
389 | 390 | _first_ip = _ips[0].strip() |
|
390 | 391 | return _first_ip |
|
391 | 392 | return ip |
|
392 | 393 | |
|
393 | 394 | |
|
394 | 395 | def _filter_port(ip): |
|
395 | 396 | """ |
|
396 | 397 | Removes a port from ip, there are 4 main cases to handle here. |
|
397 | 398 | - ipv4 eg. 127.0.0.1 |
|
398 | 399 | - ipv6 eg. ::1 |
|
399 | 400 | - ipv4+port eg. 127.0.0.1:8080 |
|
400 | 401 | - ipv6+port eg. [::1]:8080 |
|
401 | 402 | |
|
402 | 403 | :param ip: |
|
403 | 404 | """ |
|
404 | ||
|
405 | 405 | def is_ipv6(ip_addr): |
|
406 |
if hasattr(socket, |
|
|
406 | if hasattr(socket, 'inet_pton'): | |
|
407 | 407 | try: |
|
408 | 408 | socket.inet_pton(socket.AF_INET6, ip_addr) |
|
409 | 409 | except socket.error: |
|
410 | 410 | return False |
|
411 | 411 | else: |
|
412 | 412 | return False |
|
413 | 413 | return True |
|
414 | 414 | |
|
415 |
if |
|
|
415 | if ':' not in ip: # must be ipv4 pure ip | |
|
416 | 416 | return ip |
|
417 | 417 | |
|
418 |
if |
|
|
419 |
return ip.split( |
|
|
418 | if '[' in ip and ']' in ip: # ipv6 with port | |
|
419 | return ip.split(']')[0][1:].lower() | |
|
420 | 420 | |
|
421 | 421 | # must be ipv6 or ipv4 with port |
|
422 | 422 | if is_ipv6(ip): |
|
423 | 423 | return ip |
|
424 | 424 | else: |
|
425 |
ip, _port = ip.split( |
|
|
425 | ip, _port = ip.split(':')[:2] # means ipv4+port | |
|
426 | 426 | return ip |
|
427 | 427 | |
|
428 | 428 | |
|
429 | 429 | def get_ip_addr(environ): |
|
430 |
proxy_key = |
|
|
431 |
proxy_key2 = |
|
|
432 |
def_key = |
|
|
430 | proxy_key = 'HTTP_X_REAL_IP' | |
|
431 | proxy_key2 = 'HTTP_X_FORWARDED_FOR' | |
|
432 | def_key = 'REMOTE_ADDR' | |
|
433 | 433 | |
|
434 | 434 | def _filters(x): |
|
435 | 435 | return _filter_port(_filter_proxy(x)) |
|
436 | 436 | |
|
437 | 437 | ip = environ.get(proxy_key) |
|
438 | 438 | if ip: |
|
439 | 439 | return _filters(ip) |
|
440 | 440 | |
|
441 | 441 | ip = environ.get(proxy_key2) |
|
442 | 442 | if ip: |
|
443 | 443 | return _filters(ip) |
|
444 | 444 | |
|
445 |
ip = environ.get(def_key, |
|
|
445 | ip = environ.get(def_key, '0.0.0.0') | |
|
446 | 446 | return _filters(ip) |
|
447 | 447 | |
|
448 | 448 | |
|
449 | 449 | class RhodeCodeLogger(Logger): |
|
450 | 450 | """ |
|
451 | 451 | Custom Logger that allows some customization that gunicorn doesn't allow |
|
452 | 452 | """ |
|
453 | 453 | |
|
454 | 454 | datefmt = r"%Y-%m-%d %H:%M:%S" |
|
455 | 455 | |
|
456 | 456 | def __init__(self, cfg): |
|
457 | 457 | Logger.__init__(self, cfg) |
|
458 | 458 | |
|
459 | 459 | def now(self): |
|
460 | 460 | """return date in RhodeCode Log format""" |
|
461 | 461 | now = time.time() |
|
462 | 462 | msecs = int((now - int(now)) * 1000) |
|
463 |
return time.strftime(self.datefmt, time.localtime(now)) + |
|
|
463 | return time.strftime(self.datefmt, time.localtime(now)) + '.{0:03d}'.format(msecs) | |
|
464 | 464 | |
|
465 | 465 | def atoms(self, resp, req, environ, request_time): |
|
466 |
"""Gets atoms for log formatting. |
|
|
466 | """ Gets atoms for log formatting. | |
|
467 | """ | |
|
467 | 468 | status = resp.status |
|
468 | 469 | if isinstance(status, str): |
|
469 | 470 | status = status.split(None, 1)[0] |
|
470 | 471 | atoms = { |
|
471 |
|
|
|
472 |
|
|
|
473 |
|
|
|
474 |
|
|
|
475 |
|
|
|
476 | "s": status, | |
|
477 | "m": environ.get("REQUEST_METHOD"), | |
|
478 | "U": environ.get("PATH_INFO"), | |
|
479 |
|
|
|
480 |
|
|
|
481 | "b": getattr(resp, "sent", None) is not None and str(resp.sent) or "-", | |
|
482 | "B": getattr(resp, "sent", None), | |
|
483 | "f": environ.get("HTTP_REFERER", "-"), | |
|
484 | "a": environ.get("HTTP_USER_AGENT", "-"), | |
|
485 | "T": request_time.seconds, | |
|
486 | "D": (request_time.seconds * 1000000) + request_time.microseconds, | |
|
487 | "M": (request_time.seconds * 1000) + int(request_time.microseconds / 1000), | |
|
488 |
|
|
|
489 | "p": "<%s>" % os.getpid(), | |
|
472 | 'h': get_ip_addr(environ), | |
|
473 | 'l': '-', | |
|
474 | 'u': self._get_user(environ) or '-', | |
|
475 | 't': self.now(), | |
|
476 | 'r': "%s %s %s" % (environ['REQUEST_METHOD'], | |
|
477 | environ['RAW_URI'], | |
|
478 | environ["SERVER_PROTOCOL"]), | |
|
479 | 's': status, | |
|
480 | 'm': environ.get('REQUEST_METHOD'), | |
|
481 | 'U': environ.get('PATH_INFO'), | |
|
482 | 'q': environ.get('QUERY_STRING'), | |
|
483 | 'H': environ.get('SERVER_PROTOCOL'), | |
|
484 | 'b': getattr(resp, 'sent', None) is not None and str(resp.sent) or '-', | |
|
485 | 'B': getattr(resp, 'sent', None), | |
|
486 | 'f': environ.get('HTTP_REFERER', '-'), | |
|
487 | 'a': environ.get('HTTP_USER_AGENT', '-'), | |
|
488 | 'T': request_time.seconds, | |
|
489 | 'D': (request_time.seconds * 1000000) + request_time.microseconds, | |
|
490 | 'M': (request_time.seconds * 1000) + int(request_time.microseconds/1000), | |
|
491 | 'L': "%d.%06d" % (request_time.seconds, request_time.microseconds), | |
|
492 | 'p': "<%s>" % os.getpid() | |
|
490 | 493 | } |
|
491 | 494 | |
|
492 | 495 | # add request headers |
|
493 |
if hasattr(req, |
|
|
496 | if hasattr(req, 'headers'): | |
|
494 | 497 | req_headers = req.headers |
|
495 | 498 | else: |
|
496 | 499 | req_headers = req |
|
497 | 500 | |
|
498 | 501 | if hasattr(req_headers, "items"): |
|
499 | 502 | req_headers = req_headers.items() |
|
500 | 503 | |
|
501 | 504 | atoms.update({"{%s}i" % k.lower(): v for k, v in req_headers}) |
|
502 | 505 | |
|
503 | 506 | resp_headers = resp.headers |
|
504 | 507 | if hasattr(resp_headers, "items"): |
|
505 | 508 | resp_headers = resp_headers.items() |
|
506 | 509 | |
|
507 | 510 | # add response headers |
|
508 | 511 | atoms.update({"{%s}o" % k.lower(): v for k, v in resp_headers}) |
|
509 | 512 | |
|
510 | 513 | # add environ variables |
|
511 | 514 | environ_variables = environ.items() |
|
512 | 515 | atoms.update({"{%s}e" % k.lower(): v for k, v in environ_variables}) |
|
513 | 516 | |
|
514 | 517 | return atoms |
|
515 | 518 | |
|
516 | 519 | |
|
517 | 520 | logger_class = RhodeCodeLogger |
General Comments 0
You need to be logged in to leave comments.
Login now