1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875
|
import contextlib
import errno
import logging
import math
import os
import random
import signal
import socket
import sys
import time
import traceback
import warnings
from datetime import datetime, timedelta
from enum import Enum
from random import shuffle
from types import FrameType
from typing import TYPE_CHECKING, Callable, Optional, Union
from uuid import uuid4
if TYPE_CHECKING:
try:
from resource import struct_rusage
except ImportError:
pass
from redis import Redis
from redis.client import Pipeline, PubSub, PubSubWorkerThread
from contextlib import suppress
import redis.exceptions
from . import worker_registration
from .command import PUBSUB_CHANNEL_TEMPLATE, handle_command, parse_payload
from .defaults import (
DEFAULT_JOB_MONITORING_INTERVAL,
DEFAULT_LOGGING_DATE_FORMAT,
DEFAULT_LOGGING_FORMAT,
DEFAULT_MAINTENANCE_TASK_INTERVAL,
DEFAULT_RESULT_TTL,
DEFAULT_WORKER_TTL,
)
from .exceptions import (
DequeueTimeout,
DeserializationError,
InvalidJobOperation,
ShutDownImminentException,
StopRequested,
)
from .executions import Execution
from .group import Group
from .job import Job, JobStatus, Retry
from .logutils import blue, green, setup_loghandlers, yellow
from .queue import Queue
from .registry import StartedJobRegistry, clean_registries
from .scheduler import RQScheduler
from .serializers import Serializer, resolve_serializer
from .suspension import is_suspended
from .timeouts import (
HorseMonitorTimeoutException,
JobTimeoutException,
get_default_death_penalty_class,
)
from .utils import (
as_text,
compact,
decode_redis_hash,
ensure_job_list,
get_connection_from_queues,
get_version,
import_job_class,
import_queue_class,
now,
utcformat,
utcparse,
)
from .version import VERSION
try:
from setproctitle import setproctitle as setprocname
except ImportError:
def setprocname(title: str) -> None:
pass
# Set initial level to INFO.
logger = logging.getLogger('rq.worker')
if logger.level == logging.NOTSET:
logger.setLevel(logging.INFO)
_signames = dict(
(getattr(signal, signame), signame) for signame in dir(signal) if signame.startswith('SIG') and '_' not in signame
)
def signal_name(signum):
try:
if sys.version_info[:2] >= (3, 5):
return signal.Signals(signum).name
else:
return _signames[signum]
except KeyError:
return 'SIG_UNKNOWN'
except ValueError:
return 'SIG_UNKNOWN'
SHUTDOWN_SIGNAL = signal.SIGTERM if not hasattr(signal, 'SIGKILL') else signal.SIGKILL
class DequeueStrategy(str, Enum):
DEFAULT = 'default'
ROUND_ROBIN = 'round_robin'
RANDOM = 'random'
class WorkerStatus(str, Enum):
STARTED = 'started'
SUSPENDED = 'suspended'
BUSY = 'busy'
IDLE = 'idle'
class BaseWorker:
redis_worker_namespace_prefix = 'rq:worker:'
redis_workers_keys = worker_registration.REDIS_WORKER_KEYS
death_penalty_class = get_default_death_penalty_class()
queue_class = Queue
job_class = Job
# `log_result_lifespan` controls whether "Result is kept for XXX seconds"
# messages are logged after every job, by default they are.
log_result_lifespan = True
# `log_job_description` is used to toggle logging an entire jobs description.
log_job_description = True
# factor to increase connection_wait_time in case of continuous connection failures.
exponential_backoff_factor = 2.0
# Max Wait time (in seconds) after which exponential_backoff_factor won't be applicable.
max_connection_wait_time = 60.0
def __init__(
self,
queues,
name: Optional[str] = None,
default_result_ttl=DEFAULT_RESULT_TTL,
connection: Optional['Redis'] = None,
exc_handler=None,
exception_handlers=None,
maintenance_interval: int = DEFAULT_MAINTENANCE_TASK_INTERVAL,
default_worker_ttl: Optional[int] = None, # TODO remove this arg in 3.0
worker_ttl: Optional[int] = None,
job_class: Optional[Union[type[Job], str]] = None,
queue_class: Optional[Union[type[Queue], str]] = None,
log_job_description: bool = True,
job_monitoring_interval=DEFAULT_JOB_MONITORING_INTERVAL,
disable_default_exception_handler: bool = False,
prepare_for_work: bool = True,
serializer: Optional[Union[Serializer, str]] = None,
work_horse_killed_handler: Optional[Callable[[Job, int, int, 'struct_rusage'], None]] = None,
): # noqa
self.default_result_ttl = default_result_ttl
if worker_ttl:
self.worker_ttl = worker_ttl
elif default_worker_ttl:
warnings.warn('default_worker_ttl is deprecated, use worker_ttl.', DeprecationWarning, stacklevel=2)
self.worker_ttl = default_worker_ttl
else:
self.worker_ttl = DEFAULT_WORKER_TTL
self.job_monitoring_interval = job_monitoring_interval
self.maintenance_interval = maintenance_interval
if not connection:
connection = get_connection_from_queues(queues)
assert connection
connection = self._set_connection(connection)
self.connection = connection
self.redis_server_version = None
if job_class:
self.job_class = import_job_class(job_class) if isinstance(job_class, str) else job_class
if queue_class:
self.queue_class = import_queue_class(queue_class) if isinstance(queue_class, str) else queue_class
self.version: str = VERSION
self.python_version: str = sys.version
self.serializer = resolve_serializer(serializer)
self.execution: Optional[Execution] = None
queues = [
(
self.queue_class(
name=q,
connection=connection,
job_class=self.job_class,
serializer=self.serializer,
death_penalty_class=self.death_penalty_class,
)
if isinstance(q, str)
else q
)
for q in ensure_job_list(queues)
]
self.name: str = name or uuid4().hex
self.queues = queues
self.validate_queues()
self._ordered_queues = self.queues[:]
self._exc_handlers: list[Callable] = []
self._work_horse_killed_handler = work_horse_killed_handler
self._shutdown_requested_date: Optional[datetime] = None
self._state: str = 'starting'
self._is_horse: bool = False
self._horse_pid: int = 0
self._stop_requested: bool = False
self._stopped_job_id = None
self.log = logger
self.log_job_description = log_job_description
self.last_cleaned_at = None
self.successful_job_count: int = 0
self.failed_job_count: int = 0
self.total_working_time: float = 0
self.current_job_working_time: float = 0
self.birth_date = None
self.scheduler: Optional[RQScheduler] = None
self.pubsub: Optional[PubSub] = None
self.pubsub_thread = None
self._dequeue_strategy: Optional[DequeueStrategy] = DequeueStrategy.DEFAULT
self.disable_default_exception_handler = disable_default_exception_handler
if prepare_for_work:
self.hostname: Optional[str] = socket.gethostname()
self.pid: Optional[int] = os.getpid()
try:
connection.client_setname(self.name)
except redis.exceptions.ResponseError:
warnings.warn('CLIENT SETNAME command not supported, setting ip_address to unknown', Warning)
self.ip_address = 'unknown'
else:
client_addresses = [
client['addr'] for client in connection.client_list() if client.get('name') == self.name
]
if len(client_addresses) > 0:
self.ip_address = client_addresses[0]
else:
warnings.warn('CLIENT LIST command not supported, setting ip_address to unknown', Warning)
self.ip_address = 'unknown'
else:
self.hostname = None
self.pid = None
self.ip_address = 'unknown'
if isinstance(exception_handlers, (list, tuple)):
for handler in exception_handlers:
self.push_exc_handler(handler)
elif exception_handlers is not None:
self.push_exc_handler(exception_handlers)
@classmethod
def find_by_key(
cls,
worker_key: str,
connection: 'Redis',
job_class: Optional[type['Job']] = None,
queue_class: Optional[type['Queue']] = None,
serializer: Optional[Union[Serializer, str]] = None,
) -> Optional['BaseWorker']:
"""Returns a Worker instance, based on the naming conventions for
naming the internal Redis keys. Can be used to reverse-lookup Workers
by their Redis keys.
Args:
worker_key (str): The worker key
connection (Optional[Redis], optional): Redis connection. Defaults to None.
job_class (Optional[Type[Job]], optional): The job class if custom class is being used. Defaults to None.
queue_class (Optional[Type[Queue]]): The queue class if a custom class is being used. Defaults to None.
serializer (Optional[Union[Serializer, str]], optional): The serializer to use. Defaults to None.
Raises:
ValueError: If the key doesn't start with `rq:worker:`, the default worker namespace prefix.
Returns:
worker (Worker): The Worker instance.
"""
prefix = cls.redis_worker_namespace_prefix
if not worker_key.startswith(prefix):
raise ValueError('Not a valid RQ worker key: %s' % worker_key)
if not connection.exists(worker_key):
connection.srem(cls.redis_workers_keys, worker_key)
return None
name = worker_key[len(prefix) :]
worker = cls(
[],
name,
connection=connection,
job_class=job_class,
queue_class=queue_class,
prepare_for_work=False,
serializer=serializer,
)
worker.refresh()
return worker
@classmethod
def all(
cls,
connection: Optional['Redis'] = None,
job_class: Optional[type['Job']] = None,
queue_class: Optional[type['Queue']] = None,
queue: Optional['Queue'] = None,
serializer=None,
) -> list['BaseWorker']:
"""Returns an iterable of all Workers.
Returns:
workers (List[Worker]): A list of workers
"""
if queue:
connection = queue.connection
assert connection
worker_keys = worker_registration.get_keys(queue=queue, connection=connection)
workers = [
cls.find_by_key(
key, connection=connection, job_class=job_class, queue_class=queue_class, serializer=serializer
)
for key in worker_keys
]
return compact(workers)
@classmethod
def all_keys(cls, connection: Optional['Redis'] = None, queue: Optional['Queue'] = None) -> list[str]:
"""List of worker keys
Args:
connection (Optional[Redis], optional): A Redis Connection. Defaults to None.
queue (Optional[Queue], optional): The Queue. Defaults to None.
Returns:
list_keys (List[str]): A list of worker keys
"""
return [as_text(key) for key in worker_registration.get_keys(queue=queue, connection=connection)]
@classmethod
def count(cls, connection: Optional['Redis'] = None, queue: Optional['Queue'] = None) -> int:
"""Returns the number of workers by queue or connection.
Args:
connection (Optional[Redis], optional): Redis connection. Defaults to None.
queue (Optional[Queue], optional): The queue to use. Defaults to None.
Returns:
length (int): The queue length.
"""
return len(worker_registration.get_keys(queue=queue, connection=connection))
def refresh(self):
"""Refreshes the worker data.
It will get the data from the datastore and update the Worker's attributes
"""
raw_data = self.connection.hgetall(self.key)
if not raw_data:
return
data = decode_redis_hash(raw_data, decode_values=True)
self.hostname = data.get('hostname') or None
self.ip_address = data.get('ip_address') or None
self.pid = int(data['pid']) if data.get('pid') else None
self.version = data.get('version') or VERSION
self.python_version = data.get('python_version') or sys.version
self._state = data.get('state', '?')
self._job_id = data.get('current_job')
if data.get('last_heartbeat'):
self.last_heartbeat = utcparse(data['last_heartbeat'])
else:
self.last_heartbeat = None
if data.get('birth'):
self.birth_date = utcparse(data['birth'])
else:
self.birth_date = None
self.failed_job_count = int(data['failed_job_count']) if data.get('failed_job_count') else 0
self.successful_job_count = int(data['successful_job_count']) if data.get('successful_job_count') else 0
self.total_working_time = float(data['total_working_time']) if data.get('total_working_time') else 0
self.current_job_working_time = (
float(data['current_job_working_time']) if data.get('current_job_working_time') else 0
)
if data.get('queues'):
self.queues = [
self.queue_class(
queue, connection=self.connection, job_class=self.job_class, serializer=self.serializer
)
for queue in data['queues'].split(',')
]
@property
def should_run_maintenance_tasks(self):
"""Maintenance tasks should run on first startup or every 10 minutes."""
if self.last_cleaned_at is None:
return True
if (now() - self.last_cleaned_at) > timedelta(seconds=self.maintenance_interval):
return True
return False
def _set_connection(self, connection: 'Redis') -> 'Redis':
"""Configures the Redis connection's socket timeout.
This will timeout the connection in case any specific command hangs at any given time (eg. BLPOP), but
also ensures that the timeout is long enough for those operations.
If the connection provided already has an adequate `socket_timeout` defined, skips.
Args:
connection (Optional[Redis]): The Redis Connection.
"""
current_socket_timeout = connection.connection_pool.connection_kwargs.get('socket_timeout')
if current_socket_timeout is None or current_socket_timeout < self.connection_timeout:
timeout_config = {'socket_timeout': self.connection_timeout}
connection.connection_pool.connection_kwargs.update(timeout_config)
return connection
@property
def dequeue_timeout(self) -> int:
return max(1, self.worker_ttl - 15)
@property
def connection_timeout(self) -> int:
return self.dequeue_timeout + 10
def clean_registries(self):
"""Runs maintenance jobs on each Queue's registries."""
for queue in self.queues:
# If there are multiple workers running, we only want 1 worker
# to run clean_registries().
if queue.acquire_maintenance_lock():
self.log.info('Worker %s: cleaning registries for queue: %s', self.name, queue.name)
clean_registries(queue, self._exc_handlers)
worker_registration.clean_worker_registry(queue)
queue.intermediate_queue.cleanup(self, queue)
queue.release_maintenance_lock()
self.last_cleaned_at = now()
def get_redis_server_version(self):
"""Return Redis server version of connection"""
if not self.redis_server_version:
self.redis_server_version = get_version(self.connection)
return self.redis_server_version
def validate_queues(self):
"""Sanity check for the given queues."""
for queue in self.queues:
if not isinstance(queue, self.queue_class):
raise TypeError(f'{queue} is not of type {self.queue_class} or string types')
def queue_names(self) -> list[str]:
"""Returns the queue names of this worker's queues.
Returns:
List[str]: The queue names.
"""
return [queue.name for queue in self.queues]
def queue_keys(self) -> list[str]:
"""Returns the Redis keys representing this worker's queues.
Returns:
List[str]: The list of strings with queues keys
"""
return [queue.key for queue in self.queues]
@property
def key(self):
"""Returns the worker's Redis hash key."""
return self.redis_worker_namespace_prefix + self.name
@property
def pubsub_channel_name(self):
"""Returns the worker's Redis hash key."""
return PUBSUB_CHANNEL_TEMPLATE % self.name
def request_stop(self, signum, frame):
"""Stops the current worker loop but waits for child processes to
end gracefully (warm shutdown).
Args:
signum (Any): Signum
frame (Any): Frame
"""
self.log.debug('Worker %s: got signal %s', self.name, signal_name(signum))
self._shutdown_requested_date = now()
signal.signal(signal.SIGINT, self.request_force_stop)
signal.signal(signal.SIGTERM, self.request_force_stop)
self.handle_warm_shutdown_request()
self._shutdown()
def _shutdown(self):
"""
If shutdown is requested in the middle of a job, wait until
finish before shutting down and save the request in redis
"""
if self.get_state() == WorkerStatus.BUSY:
self._stop_requested = True
self.set_shutdown_requested_date()
self.log.debug(
'Worker %s: stopping after current horse is finished. Press Ctrl+C again for a cold shutdown.',
self.name,
)
if self.scheduler:
self.stop_scheduler()
else:
if self.scheduler:
self.stop_scheduler()
raise StopRequested()
def request_force_stop(self, signum: int, frame: Optional[FrameType]):
raise NotImplementedError()
def _install_signal_handlers(self):
"""Installs signal handlers for handling SIGINT and SIGTERM gracefully."""
signal.signal(signal.SIGINT, self.request_stop)
signal.signal(signal.SIGTERM, self.request_stop)
def execute_job(self, job: 'Job', queue: 'Queue'):
"""To be implemented by subclasses."""
raise NotImplementedError
def work(
self,
burst: bool = False,
logging_level: Optional[str] = None,
date_format: str = DEFAULT_LOGGING_DATE_FORMAT,
log_format: str = DEFAULT_LOGGING_FORMAT,
max_jobs: Optional[int] = None,
max_idle_time: Optional[int] = None,
with_scheduler: bool = False,
dequeue_strategy: DequeueStrategy = DequeueStrategy.DEFAULT,
) -> bool:
"""Starts the work loop.
Pops and performs all jobs on the current list of queues. When all
queues are empty, block and wait for new jobs to arrive on any of the
queues, unless `burst` mode is enabled.
If `max_idle_time` is provided, worker will die when it's idle for more than the provided value.
The return value indicates whether any jobs were processed.
Args:
burst (bool, optional): Whether to work on burst mode. Defaults to False.
logging_level (Optional[str], optional): Logging level to use.
If not provided, defaults to "INFO" unless a class-level logging level is already set.
date_format (str, optional): Date Format. Defaults to DEFAULT_LOGGING_DATE_FORMAT.
log_format (str, optional): Log Format. Defaults to DEFAULT_LOGGING_FORMAT.
max_jobs (Optional[int], optional): Max number of jobs. Defaults to None.
max_idle_time (Optional[int], optional): Max seconds for worker to be idle. Defaults to None.
with_scheduler (bool, optional): Whether to run the scheduler in a separate process. Defaults to False.
dequeue_strategy (DequeueStrategy, optional): Which strategy to use to dequeue jobs.
Defaults to DequeueStrategy.DEFAULT
Returns:
worked (bool): Will return True if any job was processed, False otherwise.
"""
self.bootstrap(logging_level, date_format, log_format)
self._dequeue_strategy = dequeue_strategy
completed_jobs = 0
if with_scheduler:
self._start_scheduler(burst, logging_level, date_format, log_format)
self._install_signal_handlers()
try:
while True:
try:
self.check_for_suspension(burst)
if self.should_run_maintenance_tasks:
self.run_maintenance_tasks()
if self._stop_requested:
self.log.info('Worker %s: stopping on request', self.name)
break
timeout = None if burst else self.dequeue_timeout
result = self.dequeue_job_and_maintain_ttl(timeout, max_idle_time)
if result is None:
if burst:
self.log.info('Worker %s: done, quitting', self.name)
elif max_idle_time is not None:
self.log.info('Worker %s: idle for %d seconds, quitting', self.name, max_idle_time)
break
job, queue = result
self.execute_job(job, queue)
self.heartbeat()
completed_jobs += 1
if max_jobs is not None:
if completed_jobs >= max_jobs:
self.log.info('Worker %s: finished executing %d jobs, quitting', self.name, completed_jobs)
break
except redis.exceptions.TimeoutError:
self.log.error('Worker %s: Redis connection timeout, quitting...', self.name)
break
except StopRequested:
break
except SystemExit:
# Cold shutdown detected
raise
except: # noqa
self.log.error('Worker %s: found an unhandled exception, quitting...', self.name, exc_info=True)
break
finally:
self.teardown()
return bool(completed_jobs)
def cleanup_execution(self, job: 'Job', pipeline: 'Pipeline'):
"""Cleans up the execution of a job.
It will remove the job execution record from the `StartedJobRegistry` and delete the Execution object.
"""
self.log.debug('Cleaning up execution of job %s', job.id)
self.set_current_job_id(None, pipeline=pipeline)
if self.execution is not None:
self.execution.delete(job=job, pipeline=pipeline)
self.execution = None
def handle_warm_shutdown_request(self):
self.log.info('Worker %s [PID %d]: warm shut down requested', self.name, self.pid)
def reorder_queues(self, reference_queue: 'Queue'):
"""Reorder the queues according to the strategy.
As this can be defined both in the `Worker` initialization or in the `work` method,
it doesn't take the strategy directly, but rather uses the private `_dequeue_strategy` attribute.
Args:
reference_queue (Union[Queue, str]): The queues to reorder
"""
if self._dequeue_strategy is None:
self._dequeue_strategy = DequeueStrategy.DEFAULT
if self._dequeue_strategy not in ('default', 'random', 'round_robin'):
raise ValueError(
f'Dequeue strategy {self._dequeue_strategy} is not allowed. Use `default`, `random` or `round_robin`.'
)
if self._dequeue_strategy == DequeueStrategy.DEFAULT:
return
if self._dequeue_strategy == DequeueStrategy.ROUND_ROBIN:
pos = self._ordered_queues.index(reference_queue)
self._ordered_queues = self._ordered_queues[pos + 1 :] + self._ordered_queues[: pos + 1]
return
if self._dequeue_strategy == DequeueStrategy.RANDOM:
shuffle(self._ordered_queues)
return
def handle_job_failure(self, job: 'Job', queue: 'Queue', started_job_registry=None, exc_string=''):
"""
Handles the failure or an executing job by:
1. Setting the job status to failed
2. Removing the job from StartedJobRegistry
3. Setting the workers current job to None
4. Add the job to FailedJobRegistry
`save_exc_to_job` should only be used for testing purposes
"""
self.log.debug('Worker %s: handling failed execution of job %s', self.name, job.id)
with self.connection.pipeline() as pipeline:
if started_job_registry is None:
started_job_registry = StartedJobRegistry(
job.origin, self.connection, job_class=self.job_class, serializer=self.serializer
)
# check whether a job was stopped intentionally and set the job
# status appropriately if it was this job.
job_is_stopped = self._stopped_job_id == job.id
retry = job.should_retry and not job_is_stopped
if job_is_stopped:
job.set_status(JobStatus.STOPPED, pipeline=pipeline)
self._stopped_job_id = None
else:
# Requeue/reschedule if retry is configured, otherwise
if not retry:
job.set_status(JobStatus.FAILED, pipeline=pipeline)
self.cleanup_execution(job, pipeline=pipeline)
if not self.disable_default_exception_handler and not retry:
job._handle_failure(exc_string, pipeline=pipeline, worker_name=self.name)
with suppress(redis.exceptions.ConnectionError):
pipeline.execute()
self.increment_failed_job_count(pipeline)
if job.started_at and job.ended_at:
self.increment_total_working_time(job.ended_at - job.started_at, pipeline)
if retry:
job.retry(queue, pipeline)
enqueue_dependents = False
else:
enqueue_dependents = True
try:
pipeline.execute()
if enqueue_dependents:
queue.enqueue_dependents(job)
except Exception as e:
# Ensure that custom exception handlers are called
# even if Redis is down
self.log.error(
'Worker %s: exception during pipeline execute or enqueue_dependents for job %s: %s',
self.name,
job.id,
e,
)
pass
def set_current_job_working_time(self, current_job_working_time: float, pipeline: Optional['Pipeline'] = None):
"""Sets the current job working time in seconds
Args:
current_job_working_time (float): The current job working time in seconds
pipeline (Optional[Pipeline], optional): Pipeline to use. Defaults to None.
"""
self.current_job_working_time = current_job_working_time
connection = pipeline if pipeline is not None else self.connection
connection.hset(self.key, 'current_job_working_time', current_job_working_time)
def set_current_job_id(self, job_id: Optional[str] = None, pipeline: Optional['Pipeline'] = None):
"""Sets the current job id.
If `None` is used it will delete the current job key.
Args:
job_id (Optional[str], optional): The job id. Defaults to None.
pipeline (Optional[Pipeline], optional): The pipeline to use. Defaults to None.
"""
connection = pipeline if pipeline is not None else self.connection
if job_id is None:
connection.hdel(self.key, 'current_job')
else:
connection.hset(self.key, 'current_job', job_id)
def get_current_job_id(self, pipeline: Optional['Pipeline'] = None) -> Optional[str]:
"""Retrieves the current job id.
Args:
pipeline (Optional['Pipeline'], optional): The pipeline to use. Defaults to None.
Returns:
job_id (Optional[str): The job id
"""
connection = pipeline if pipeline is not None else self.connection
result = connection.hget(self.key, 'current_job')
if result is None:
return None
return as_text(result)
def get_current_job(self) -> Optional['Job']:
"""Returns the currently executing job instance.
Returns:
job (Job): The job instance.
"""
job_id = self.get_current_job_id()
if job_id is None:
return None
return self.job_class.fetch(job_id, self.connection, self.serializer)
def set_state(self, state: str, pipeline: Optional['Pipeline'] = None):
"""Sets the worker's state.
Args:
state (str): The state
pipeline (Optional[Pipeline], optional): The pipeline to use. Defaults to None.
"""
self._state = state
connection = pipeline if pipeline is not None else self.connection
connection.hset(self.key, 'state', state)
def _set_state(self, state):
"""Raise a DeprecationWarning if ``worker.state = X`` is used"""
warnings.warn('worker.state is deprecated, use worker.set_state() instead.', DeprecationWarning)
self.set_state(state)
def get_state(self) -> str:
return self._state
def _get_state(self):
"""Raise a DeprecationWarning if ``worker.state == X`` is used"""
warnings.warn('worker.state is deprecated, use worker.get_state() instead.', DeprecationWarning)
return self.get_state()
state = property(_get_state, _set_state)
def _start_scheduler(
self,
burst: bool = False,
logging_level: Optional[str] = 'INFO',
date_format: str = DEFAULT_LOGGING_DATE_FORMAT,
log_format: str = DEFAULT_LOGGING_FORMAT,
):
"""Starts the scheduler process.
This is specifically designed to be run by the worker when running the `work()` method.
Instantiates the RQScheduler and tries to acquire a lock.
If the lock is acquired, start scheduler.
If worker is on burst mode just enqueues scheduled jobs and quits,
otherwise, starts the scheduler in a separate process.
Args:
burst (bool, optional): Whether to work on burst mode. Defaults to False.
logging_level (str, optional): Logging level to use. Defaults to "INFO".
date_format (str, optional): Date Format. Defaults to DEFAULT_LOGGING_DATE_FORMAT.
log_format (str, optional): Log Format. Defaults to DEFAULT_LOGGING_FORMAT.
"""
self.scheduler = RQScheduler(
self.queues,
connection=self.connection,
logging_level=logging_level if logging_level is not None else self.log.level,
date_format=date_format,
log_format=log_format,
serializer=self.serializer,
)
self.scheduler.acquire_locks()
if self.scheduler.acquired_locks:
if burst:
self.scheduler.enqueue_scheduled_jobs()
self.scheduler.release_locks()
else:
self.scheduler.start()
def register_birth(self):
"""Registers its own birth."""
self.log.debug('Worker %s: registering birth', self.name)
if self.connection.exists(self.key) and not self.connection.hexists(self.key, 'death'):
msg = 'There exists an active worker named {0!r} already'
raise ValueError(msg.format(self.name))
key = self.key
queues = ','.join(self.queue_names())
with self.connection.pipeline() as p:
p.delete(key)
right_now = now()
now_in_string = utcformat(right_now)
self.birth_date = right_now
mapping = {
'birth': now_in_string,
'last_heartbeat': now_in_string,
'queues': queues,
'pid': self.pid,
'hostname': self.hostname,
'ip_address': self.ip_address,
'version': self.version,
'python_version': self.python_version,
}
p.hset(key, mapping=mapping)
worker_registration.register(self, p)
p.expire(key, self.worker_ttl + 60)
p.execute()
def register_death(self):
"""Registers its own death."""
self.log.debug('Worker %s: registering death', self.name)
with self.connection.pipeline() as p:
# We cannot use self.state = 'dead' here, because that would
# rollback the pipeline
worker_registration.unregister(self, p)
p.hset(self.key, 'death', utcformat(now()))
p.expire(self.key, 60)
p.execute()
@property
def horse_pid(self):
"""The horse's process ID. Only available in the worker. Will return
0 in the horse part of the fork.
"""
return self._horse_pid
def bootstrap(
self,
logging_level: Optional[str] = 'INFO',
date_format: str = DEFAULT_LOGGING_DATE_FORMAT,
log_format: str = DEFAULT_LOGGING_FORMAT,
):
"""Bootstraps the worker.
Runs the basic tasks that should run when the worker actually starts working.
Used so that new workers can focus on the work loop implementation rather
than the full bootstrapping process.
Args:
logging_level (str, optional): Logging level to use. Defaults to "INFO".
date_format (str, optional): Date Format. Defaults to DEFAULT_LOGGING_DATE_FORMAT.
log_format (str, optional): Log Format. Defaults to DEFAULT_LOGGING_FORMAT.
"""
setup_loghandlers(logging_level, date_format, log_format, name='rq.worker')
setup_loghandlers(logging_level, date_format, log_format, name='rq.job')
self.register_birth()
self.log.info('Worker %s: started with PID %d, version %s', self.name, os.getpid(), VERSION)
self.subscribe()
self.set_state(WorkerStatus.STARTED)
qnames = self.queue_names()
self.log.info('*** Listening on %s...', green(', '.join(qnames)))
def check_for_suspension(self, burst: bool):
"""Check to see if workers have been suspended by `rq suspend`"""
before_state = None
notified = False
while not self._stop_requested and is_suspended(self.connection, self):
if burst:
self.log.info('Worker %s: suspended in burst mode, exiting', self.name)
self.log.info('Worker %s: note: there could still be unfinished jobs on the queue', self.name)
raise StopRequested
if not notified:
self.log.info('Worker %s: suspended, run `rq resume` to resume', self.name)
before_state = self.get_state()
self.set_state(WorkerStatus.SUSPENDED)
notified = True
time.sleep(1)
if before_state:
self.set_state(before_state)
def procline(self, message):
"""Changes the current procname for the process.
This can be used to make `ps -ef` output more readable.
"""
setprocname(f'rq:worker:{self.name}: {message}')
def set_shutdown_requested_date(self):
"""Sets the date on which the worker received a (warm) shutdown request"""
self.connection.hset(self.key, 'shutdown_requested_date', utcformat(self._shutdown_requested_date))
@property
def shutdown_requested_date(self):
"""Fetches shutdown_requested_date from Redis."""
shutdown_requested_timestamp = self.connection.hget(self.key, 'shutdown_requested_date')
if shutdown_requested_timestamp is not None:
return utcparse(as_text(shutdown_requested_timestamp))
@property
def death_date(self):
"""Fetches death date from Redis."""
death_timestamp = self.connection.hget(self.key, 'death')
if death_timestamp is not None:
return utcparse(as_text(death_timestamp))
def run_maintenance_tasks(self):
"""
Runs periodic maintenance tasks, these include:
1. Check if scheduler should be started. This check should not be run
on first run since worker.work() already calls
`scheduler.enqueue_scheduled_jobs()` on startup.
2. Cleaning registries
No need to try to start scheduler on first run
"""
if self.last_cleaned_at:
if self.scheduler and (not self.scheduler._process or not self.scheduler._process.is_alive()):
self.scheduler.acquire_locks(auto_start=True)
self.clean_registries()
Group.clean_registries(connection=self.connection)
def _pubsub_exception_handler(self, exc: Exception, pubsub: 'PubSub', pubsub_thread: 'PubSubWorkerThread') -> None:
"""
This exception handler allows the pubsub_thread to continue & retry to
connect after a connection problem the same way the main worker loop
indefinitely retries.
redis-py internal mechanism will restore the channels subscriptions
once the connection is re-established.
"""
if isinstance(exc, (redis.exceptions.ConnectionError)):
self.log.error(
'Worker %s: could not connect to Redis instance: %s retrying in %d seconds...',
self.name,
exc,
2,
)
time.sleep(2.0)
else:
self.log.warning('Worker %s: pubsub thread exiting on %s', self.name, exc)
raise
def handle_payload(self, message):
"""Handle external commands"""
self.log.debug('Worker %s: received message: %s', self.name, message)
payload = parse_payload(message)
handle_command(self, payload)
def subscribe(self):
"""Subscribe to this worker's channel"""
self.log.info('Worker %s: subscribing to channel %s', self.name, self.pubsub_channel_name)
self.pubsub = self.connection.pubsub()
self.pubsub.subscribe(**{self.pubsub_channel_name: self.handle_payload})
self.pubsub_thread = self.pubsub.run_in_thread(
sleep_time=60, daemon=True, exception_handler=self._pubsub_exception_handler
)
def get_heartbeat_ttl(self, job: 'Job') -> int:
"""Get's the TTL for the next heartbeat.
Args:
job (Job): The Job
Returns:
int: The heartbeat TTL.
"""
if job.timeout and job.timeout > 0:
remaining_execution_time = job.timeout - self.current_job_working_time
return int(min(remaining_execution_time, self.job_monitoring_interval)) + 60
else:
return self.job_monitoring_interval + 60
def prepare_execution(self, job: 'Job') -> Execution:
"""This method is called by the main `Worker` (not the horse) as it prepares for execution.
Do not confuse this with worker.prepare_job_execution() which is called by the horse.
"""
with self.connection.pipeline() as pipeline:
heartbeat_ttl = self.get_heartbeat_ttl(job)
self.execution = Execution.create(job, heartbeat_ttl, pipeline=pipeline)
self.set_state(WorkerStatus.BUSY, pipeline=pipeline)
pipeline.execute()
return self.execution
def unsubscribe(self):
"""Unsubscribe from pubsub channel"""
if self.pubsub_thread:
self.log.info('Worker %s: unsubscribing from channel %s', self.name, self.pubsub_channel_name)
self.pubsub.unsubscribe()
self.pubsub_thread.stop()
self.pubsub_thread.join(timeout=1)
self.pubsub.close()
def dequeue_job_and_maintain_ttl(
self, timeout: Optional[int], max_idle_time: Optional[int] = None
) -> Optional[tuple['Job', 'Queue']]:
"""Dequeues a job while maintaining the TTL.
Returns:
result (Tuple[Job, Queue]): A tuple with the job and the queue.
"""
result = None
qnames = ','.join(self.queue_names())
self.set_state(WorkerStatus.IDLE)
self.procline('Listening on ' + qnames)
self.log.debug('Worker %s: *** Listening on %s...', self.name, green(qnames))
connection_wait_time = 1.0
idle_since = now()
idle_time_left = max_idle_time
while True:
try:
self.heartbeat()
if self.should_run_maintenance_tasks:
self.run_maintenance_tasks()
if timeout is not None and idle_time_left is not None:
timeout = min(timeout, idle_time_left)
self.log.debug(
'Worker %s: dequeueing jobs on queues %s and timeout %s', self.name, green(qnames), timeout
)
result = self.queue_class.dequeue_any(
self._ordered_queues,
timeout,
connection=self.connection,
job_class=self.job_class,
serializer=self.serializer,
death_penalty_class=self.death_penalty_class,
)
if result is not None:
job, queue = result
self.reorder_queues(reference_queue=queue)
self.log.debug('Worker %s: dequeued job %s from %s', self.name, blue(job.id), green(queue.name))
job.redis_server_version = self.get_redis_server_version()
if self.log_job_description:
self.log.info('%s: %s (%s)', green(queue.name), blue(job.description), job.id)
else:
self.log.info('%s: %s', green(queue.name), job.id)
break
except DequeueTimeout:
if max_idle_time is not None:
idle_for = (now() - idle_since).total_seconds()
idle_time_left = math.ceil(max_idle_time - idle_for)
if idle_time_left <= 0:
break
except redis.exceptions.ConnectionError as conn_err:
self.log.error(
'Worker %s: could not connect to Redis instance: %s retrying in %d seconds...',
self.name,
conn_err,
connection_wait_time,
)
time.sleep(connection_wait_time)
connection_wait_time *= self.exponential_backoff_factor
connection_wait_time = min(connection_wait_time, self.max_connection_wait_time)
self.heartbeat()
return result
def heartbeat(self, timeout: Optional[int] = None, pipeline: Optional['Pipeline'] = None):
"""Specifies a new worker timeout, typically by extending the
expiration time of the worker, effectively making this a "heartbeat"
to not expire the worker until the timeout passes.
The next heartbeat should come before this time, or the worker will
die (at least from the monitoring dashboards).
If no timeout is given, the worker_ttl will be used to update
the expiration time of the worker.
Args:
timeout (Optional[int]): Timeout
pipeline (Optional[Redis]): A Redis pipeline
"""
timeout = timeout or self.worker_ttl + 60
connection: Union[Redis, Pipeline] = pipeline if pipeline is not None else self.connection
connection.expire(self.key, timeout)
connection.hset(self.key, 'last_heartbeat', utcformat(now()))
self.log.debug(
'Worker %s: sent heartbeat to prevent worker timeout. Next one should arrive in %s seconds.',
self.name,
timeout,
)
def teardown(self):
if not self.is_horse:
if self.scheduler:
self.stop_scheduler()
self.register_death()
self.unsubscribe()
def stop_scheduler(self):
"""Ensure scheduler process is stopped
Will send the kill signal to scheduler process,
if there's an OSError, just passes and `join()`'s the scheduler process,
waiting for the process to finish.
"""
if self.scheduler._process and self.scheduler._process.pid:
try:
os.kill(self.scheduler._process.pid, signal.SIGTERM)
except OSError:
pass
self.scheduler._process.join()
def increment_failed_job_count(self, pipeline: Optional['Pipeline'] = None):
"""Used to keep the worker stats up to date in Redis.
Increments the failed job count.
Args:
pipeline (Optional[Pipeline], optional): A Redis Pipeline. Defaults to None.
"""
connection = pipeline if pipeline is not None else self.connection
connection.hincrby(self.key, 'failed_job_count', 1)
def increment_successful_job_count(self, pipeline: Optional['Pipeline'] = None):
"""Used to keep the worker stats up to date in Redis.
Increments the successful job count.
Args:
pipeline (Optional[Pipeline], optional): A Redis Pipeline. Defaults to None.
"""
connection = pipeline if pipeline is not None else self.connection
connection.hincrby(self.key, 'successful_job_count', 1)
def increment_total_working_time(self, job_execution_time: timedelta, pipeline: 'Pipeline'):
"""Used to keep the worker stats up to date in Redis.
Increments the time the worker has been working for (in seconds).
Args:
job_execution_time (timedelta): A timedelta object.
pipeline (Optional[Pipeline], optional): A Redis Pipeline. Defaults to None.
"""
pipeline.hincrbyfloat(self.key, 'total_working_time', job_execution_time.total_seconds())
def handle_exception(self, job: 'Job', *exc_info):
"""Walks the exception handler stack to delegate exception handling.
If the job cannot be deserialized, it will raise when func_name or
the other properties are accessed, which will stop exceptions from
being properly logged, so we guard against it here.
"""
self.log.debug('Worker %s: handling exception for %s.', self.name, job.id)
exc_string = ''.join(traceback.format_exception(*exc_info))
try:
extra = {'func': job.func_name, 'arguments': job.args, 'kwargs': job.kwargs}
func_name = job.func_name
except DeserializationError:
extra = {}
func_name = '<DeserializationError>'
# the properties below should be safe however
extra.update({'queue': job.origin, 'job_id': job.id})
# func_name
self.log.error(
'Worker %s: job %s: exception raised while executing (%s)\n%s',
self.name,
job.id,
func_name,
exc_string,
extra=extra,
)
for handler in self._exc_handlers:
self.log.debug('Worker %s: invoking exception handler %s', self.name, handler)
fallthrough = handler(job, *exc_info)
# Only handlers with explicit return values should disable further
# exc handling, so interpret a None return value as True.
if fallthrough is None:
fallthrough = True
if not fallthrough:
break
def push_exc_handler(self, handler_func):
"""Pushes an exception handler onto the exc handler stack."""
self._exc_handlers.append(handler_func)
def kill_horse(self, sig: signal.Signals = SHUTDOWN_SIGNAL):
raise NotImplementedError()
class Worker(BaseWorker):
@property
def is_horse(self):
"""Returns whether or not this is the worker or the work horse."""
return self._is_horse
def kill_horse(self, sig: signal.Signals = SHUTDOWN_SIGNAL):
"""Kill the horse but catch "No such process" error has the horse could already be dead.
Args:
sig (signal.Signals, optional): _description_. Defaults to SIGKILL.
"""
try:
os.killpg(os.getpgid(self.horse_pid), sig)
self.log.info('Worker %s: killed horse pid %s', self.name, self.horse_pid)
except OSError as e:
if e.errno == errno.ESRCH:
# "No such process" is fine with us
self.log.debug('Worker %s: horse already dead', self.name)
else:
raise
def wait_for_horse(self) -> tuple[Optional[int], Optional[int], Optional['struct_rusage']]:
"""Waits for the horse process to complete.
Uses `0` as argument as to include "any child in the process group of the current process".
"""
pid = stat = rusage = None
with contextlib.suppress(ChildProcessError): # ChildProcessError: [Errno 10] No child processes
pid, stat, rusage = os.wait4(self.horse_pid, 0)
return pid, stat, rusage
def request_force_stop(self, signum: int, frame: Optional[FrameType]):
"""Terminates the application (cold shutdown).
Args:
signum (Any): Signum
frame (Any): Frame
Raises:
SystemExit: SystemExit
"""
# When worker is run through a worker pool, it may receive duplicate signals
# One is sent by the pool when it calls `pool.stop_worker()` and another is sent by the OS
# when user hits Ctrl+C. In this case if we receive the second signal within 1 second,
# we ignore it.
if (now() - self._shutdown_requested_date) < timedelta(seconds=1): # type: ignore
self.log.debug('Worker %s: shutdown signal ignored, received twice in less than 1 second', self.name)
return
self.log.warning('Worker %s: cold shut down', self.name)
# Take down the horse with the worker
if self.horse_pid:
self.log.debug('Worker %s: taking down horse %s with me', self.name, self.horse_pid)
self.kill_horse()
self.wait_for_horse()
raise SystemExit()
def fork_work_horse(self, job: 'Job', queue: 'Queue'):
"""Spawns a work horse to perform the actual work and passes it a job.
This is where the `fork()` actually happens.
Args:
job (Job): The Job that will be ran
queue (Queue): The queue
"""
child_pid = os.fork()
os.environ['RQ_WORKER_ID'] = self.name
os.environ['RQ_JOB_ID'] = job.id
if child_pid == 0:
os.setpgrp()
self.main_work_horse(job, queue)
os._exit(0) # just in case
else:
self._horse_pid = child_pid
self.procline(f'Forked {child_pid} at {time.time()}')
def get_heartbeat_ttl(self, job: 'Job') -> int:
"""Get's the TTL for the next heartbeat.
Args:
job (Job): The Job
Returns:
int: The heartbeat TTL.
"""
if job.timeout and job.timeout > 0:
remaining_execution_time = job.timeout - self.current_job_working_time
return int(min(remaining_execution_time, self.job_monitoring_interval)) + 60
else:
return self.job_monitoring_interval + 60
def monitor_work_horse(self, job: 'Job', queue: 'Queue'):
"""The worker will monitor the work horse and make sure that it
either executes successfully or the status of the job is set to
failed
Args:
job (Job): _description_
queue (Queue): _description_
"""
retpid = ret_val = rusage = None
job.started_at = now()
while True:
try:
with self.death_penalty_class(self.job_monitoring_interval, HorseMonitorTimeoutException):
retpid, ret_val, rusage = self.wait_for_horse()
break
except HorseMonitorTimeoutException:
# Horse has not exited yet and is still running.
# Send a heartbeat to keep the worker alive.
self.set_current_job_working_time((now() - job.started_at).total_seconds())
# Kill the job from this side if something is really wrong (interpreter lock/etc).
if job.timeout != -1 and self.current_job_working_time > (job.timeout + 60): # type: ignore
self.heartbeat(self.job_monitoring_interval + 60)
self.kill_horse()
self.wait_for_horse()
break
self.maintain_heartbeats(job)
except OSError as e:
# In case we encountered an OSError due to EINTR (which is
# caused by a SIGINT or SIGTERM signal during
# os.waitpid()), we simply ignore it and enter the next
# iteration of the loop, waiting for the child to end. In
# any other case, this is some other unexpected OS error,
# which we don't want to catch, so we re-raise those ones.
if e.errno != errno.EINTR:
raise
# Send a heartbeat to keep the worker alive.
self.heartbeat()
self.set_current_job_working_time(0)
self._horse_pid = 0 # Set horse PID to 0, horse has finished working
self.log.debug(
'Worker %s: work horse finished for job %s: retpid=%s, ret_val=%s', self.name, job.id, retpid, ret_val
)
if ret_val == os.EX_OK: # The process exited normally.
return
try:
job_status = job.get_status()
except InvalidJobOperation:
return # Job completed and its ttl has expired
if self._stopped_job_id == job.id:
# Work-horse killed deliberately
self.log.warning('Worker %s: job %s stopped by user, moving job to FailedJobRegistry', self.name, job.id)
if job.stopped_callback:
job.execute_stopped_callback(self.death_penalty_class)
self.handle_job_failure(job, queue=queue, exc_string='Job stopped by user, work-horse terminated.')
elif job_status not in [JobStatus.FINISHED, JobStatus.FAILED]:
if not job.ended_at:
job.ended_at = now()
# Unhandled failure: move the job to the failed queue
signal_msg = f' (signal {os.WTERMSIG(ret_val)})' if ret_val and os.WIFSIGNALED(ret_val) else ''
exc_string = f'Work-horse terminated unexpectedly; waitpid returned {ret_val}{signal_msg}; '
self.log.warning('Worker %s: moving job %s to FailedJobRegistry (%s)', self.name, job.id, exc_string)
self.handle_work_horse_killed(job, retpid, ret_val, rusage)
self.handle_job_failure(job, queue=queue, exc_string=exc_string)
def execute_job(self, job: 'Job', queue: 'Queue'):
"""Spawns a work horse to perform the actual work and passes it a job.
The worker will wait for the work horse and make sure it executes
within the given timeout bounds, or will end the work horse with
SIGALRM.
"""
self.prepare_execution(job)
self.fork_work_horse(job, queue)
self.monitor_work_horse(job, queue)
self.set_state(WorkerStatus.IDLE)
def maintain_heartbeats(self, job: 'Job'):
"""Updates worker, execution and job's last heartbeat fields."""
with self.connection.pipeline() as pipeline:
self.heartbeat(self.job_monitoring_interval + 60, pipeline=pipeline)
ttl = int(self.get_heartbeat_ttl(job))
# Also need to update execution's heartbeat
self.execution.heartbeat(job.started_job_registry, ttl, pipeline=pipeline) # type: ignore
# After transition to job execution is complete, `job.heartbeat()` is no longer needed
job.heartbeat(now(), ttl, pipeline=pipeline, xx=True)
results = pipeline.execute()
# If job was enqueued with `result_ttl=0` (job is deleted as soon as it finishes),
# a race condition could happen where heartbeat arrives after job has been deleted,
# leaving a job key that contains only `last_heartbeat` field.
# job.heartbeat() uses hset() to update job's timestamp. This command returns 1 if a new
# Redis key is created, 0 otherwise. So in this case we check the return of job's
# heartbeat() command. If a new key was created, this means the job was already
# deleted. In this case, we simply send another delete command to remove the key.
# https://github.com/rq/rq/issues/1450
if results[7] == 1:
self.connection.delete(job.key)
def main_work_horse(self, job: 'Job', queue: 'Queue'):
"""This is the entry point of the newly spawned work horse.
After fork()'ing, always assure we are generating random sequences
that are different from the worker.
os._exit() is the way to exit from childs after a fork(), in
contrast to the regular sys.exit()
"""
random.seed()
self.setup_work_horse_signals()
self._is_horse = True
self.log = logger
try:
self.perform_job(job, queue)
except: # noqa
os._exit(1)
os._exit(0)
def setup_work_horse_signals(self):
"""Setup signal handing for the newly spawned work horse
Always ignore Ctrl+C in the work horse, as it might abort the
currently running job.
The main worker catches the Ctrl+C and requests graceful shutdown
after the current work is done. When cold shutdown is requested, it
kills the current job anyway.
"""
signal.signal(signal.SIGINT, signal.SIG_IGN)
signal.signal(signal.SIGTERM, signal.SIG_DFL)
def prepare_job_execution(self, job: 'Job', remove_from_intermediate_queue: bool = False) -> None:
"""Performs misc bookkeeping like updating states prior to
job execution.
"""
self.log.debug('Worker %s: preparing for execution of job ID %s', self.name, job.id)
with self.connection.pipeline() as pipeline:
self.set_current_job_id(job.id, pipeline=pipeline)
self.set_current_job_working_time(0, pipeline=pipeline)
heartbeat_ttl = self.get_heartbeat_ttl(job)
self.heartbeat(heartbeat_ttl, pipeline=pipeline)
job.heartbeat(now(), heartbeat_ttl, pipeline=pipeline)
job.prepare_for_execution(self.name, pipeline=pipeline)
if remove_from_intermediate_queue:
from .queue import Queue
queue = Queue(job.origin, connection=self.connection)
pipeline.lrem(queue.intermediate_queue_key, 1, job.id)
pipeline.execute()
self.log.debug('Worker %s: job preparation finished.', self.name)
msg = 'Processing {0} from {1} since {2}'
self.procline(msg.format(job.func_name, job.origin, time.time()))
def handle_job_retry(self, job: 'Job', queue: 'Queue', retry: Retry, started_job_registry: StartedJobRegistry):
"""Handles the retry of certain job.
It will remove the job from the `StartedJobRegistry` and requeue or reschedule the job.
Args:
job (Job): The job that will be retried.
queue (Queue): The queue
started_job_registry (StartedJobRegistry): The started registry
"""
self.log.debug('Worker %s: handling retry of job %s', self.name, job.id)
# Check if job has exceeded max retries
if job.number_of_retries and job.number_of_retries >= retry.max:
# If max retries exceeded, treat as failure
self.log.warning('Worker %s: job %s has exceeded maximum retry attempts (%d)', self.name, job.id, retry.max)
exc_string = f'Job failed after {retry.max} retry attempts'
self.handle_job_failure(job, queue=queue, exc_string=exc_string)
return
# Calculate retry interval based on retry count
retry_interval = Retry.get_interval(job.number_of_retries or 0, retry.intervals)
with self.connection.pipeline() as pipeline:
self.increment_failed_job_count(pipeline=pipeline)
self.increment_total_working_time(job.ended_at - job.started_at, pipeline) # type: ignore
if retry_interval > 0:
# Schedule job for later if there's an interval
scheduled_time = now() + timedelta(seconds=retry_interval)
job.set_status(JobStatus.SCHEDULED, pipeline=pipeline)
queue.schedule_job(job, scheduled_time, pipeline=pipeline)
self.log.debug(
'Worker %s: job %s: scheduled for retry at %s, %s attempts remaining',
self.name,
job.id,
scheduled_time,
retry.max - (job.number_of_retries or 0),
)
else:
self.log.debug(
'Worker %s: job %s: enqueued for retry, %s attempts remaining',
self.name,
job.id,
retry.max - (job.number_of_retries or 0),
)
job._handle_retry_result(queue=queue, pipeline=pipeline, worker_name=self.name)
self.cleanup_execution(job, pipeline=pipeline)
pipeline.execute()
self.log.debug('Worker %s: finished handling retry of job %s', self.name, job.id)
def handle_job_success(self, job: 'Job', queue: 'Queue', started_job_registry: StartedJobRegistry):
"""Handles the successful execution of certain job.
It will remove the job from the `StartedJobRegistry`, adding it to the `SuccessfulJobRegistry`,
and run a few maintenance tasks including:
- Resting the current job ID
- Enqueue dependents
- Incrementing the job count and working time
- Handling of the job successful execution
- If job.repeats_left > 0, it will be scheduled for the next execution.
Runs within a loop with the `watch` method so that protects interactions
with dependents keys.
Args:
job (Job): The job that was successful.
queue (Queue): The queue
started_job_registry (StartedJobRegistry): The started registry
"""
self.log.debug('Worker %s: handling successful execution of job %s', self.name, job.id)
with self.connection.pipeline() as pipeline:
while True:
try:
# if dependencies are inserted after enqueue_dependents
# a WatchError is thrown by execute()
pipeline.watch(job.dependents_key)
# enqueue_dependents might call multi() on the pipeline
self.log.debug('Worker %s: enqueueing dependents of job %s', self.name, job.id)
queue.enqueue_dependents(job, pipeline=pipeline)
if not pipeline.explicit_transaction:
# enqueue_dependents didn't call multi after all!
# We have to do it ourselves to make sure everything runs in a transaction
self.log.debug('Worker %s: calling multi() on pipeline for job %s', self.name, job.id)
pipeline.multi()
self.increment_successful_job_count(pipeline=pipeline)
self.increment_total_working_time(job.ended_at - job.started_at, pipeline) # type: ignore
result_ttl = job.get_result_ttl(self.default_result_ttl)
if result_ttl != 0:
self.log.debug("Worker %s: saving job %s's successful execution result", self.name, job.id)
job._handle_success(result_ttl, pipeline=pipeline, worker_name=self.name)
if job.repeats_left is not None and job.repeats_left > 0:
from .repeat import Repeat
self.log.info(
'Worker %s: job %s scheduled to repeat (%s left)', self.name, job.id, job.repeats_left
)
Repeat.schedule(job, queue, pipeline=pipeline)
else:
job.cleanup(result_ttl, pipeline=pipeline, remove_from_queue=False)
self.log.debug('Cleaning up execution of job %s', job.id)
self.cleanup_execution(job, pipeline=pipeline)
pipeline.execute()
assert job.started_at
assert job.ended_at
time_taken = job.ended_at - job.started_at
if self.log_job_description:
self.log.info(
'Successfully completed %s job in %ss on worker %s', job.description, time_taken, self.name
)
else:
self.log.info(
'Successfully completed job %s in %ss on worker %s', job.id, time_taken, self.name
)
self.log.debug('Worker %s: finished handling successful execution of job %s', self.name, job.id)
break
except redis.exceptions.WatchError:
continue
def handle_execution_ended(self, job: 'Job', queue: 'Queue', heartbeat_ttl: int):
"""Called after job has finished execution."""
job.ended_at = now()
job.heartbeat(now(), heartbeat_ttl)
def perform_job(self, job: 'Job', queue: 'Queue') -> bool:
"""Performs the actual work of a job. Will/should only be called
inside the work horse's process.
Args:
job (Job): The Job
queue (Queue): The Queue
Returns:
bool: True after finished.
"""
started_job_registry = queue.started_job_registry
self.log.debug('Worker %s: started job registry set.', self.name)
try:
remove_from_intermediate_queue = len(self.queues) == 1
self.prepare_job_execution(job, remove_from_intermediate_queue)
job.started_at = now()
timeout = job.timeout or self.queue_class.DEFAULT_TIMEOUT
with self.death_penalty_class(timeout, JobTimeoutException, job_id=job.id):
self.log.debug('Worker %s: performing job %s ...', self.name, job.id)
return_value = job.perform()
self.log.debug('Worker %s: finished performing job %s', self.name, job.id)
self.handle_execution_ended(job, queue, job.success_callback_timeout)
# Pickle the result in the same try-except block since we need
# to use the same exc handling when pickling fails
job._result = return_value
if isinstance(return_value, Retry):
# Retry the job
self.log.debug('Worker %s: job %s returns a Retry object', self.name, job.id)
self.handle_job_retry(
job=job, queue=queue, retry=return_value, started_job_registry=started_job_registry
)
return True
else:
job.execute_success_callback(self.death_penalty_class, return_value)
self.handle_job_success(job=job, queue=queue, started_job_registry=started_job_registry)
except: # NOQA
self.log.debug('Worker %s: job %s raised an exception.', self.name, job.id)
job._status = JobStatus.FAILED
self.handle_execution_ended(job, queue, job.failure_callback_timeout)
exc_info = sys.exc_info()
exc_string = ''.join(traceback.format_exception(*exc_info))
try:
job.execute_failure_callback(self.death_penalty_class, *exc_info)
except: # noqa
exc_info = sys.exc_info()
exc_string = ''.join(traceback.format_exception(*exc_info))
# TODO: reversing the order of handle_job_failure() and handle_exception()
# causes Sentry test to fail
self.handle_exception(job, *exc_info)
self.handle_job_failure(
job=job, exc_string=exc_string, queue=queue, started_job_registry=started_job_registry
)
return False
self.log.info('%s: %s (%s)', green(job.origin), blue('Job OK'), job.id)
if return_value is not None:
self.log.debug('Worker %s: result: %r', self.name, yellow(str(return_value)))
if self.log_result_lifespan:
result_ttl = job.get_result_ttl(self.default_result_ttl)
if result_ttl == 0:
self.log.info('Result discarded immediately')
elif result_ttl > 0:
self.log.info('Result is kept for %s seconds', result_ttl)
else:
self.log.info('Result will never expire, clean up result key manually')
return True
def pop_exc_handler(self):
"""Pops the latest exception handler off of the exc handler stack."""
return self._exc_handlers.pop()
def handle_work_horse_killed(self, job, retpid, ret_val, rusage):
self.log.warning('Work horse killed for job %s: retpid=%s, ret_val=%s', job.id, retpid, ret_val)
if self._work_horse_killed_handler is None:
return
self._work_horse_killed_handler(job, retpid, ret_val, rusage)
def __eq__(self, other):
"""Equality does not take the database/connection into account"""
if not isinstance(other, self.__class__):
raise TypeError('Cannot compare workers to other types (of workers)')
return self.name == other.name
def __hash__(self):
"""The hash does not take the database/connection into account"""
return hash(self.name)
class SpawnWorker(Worker):
"""Worker implementation that uses os.spawn() instead of os.fork().
This implementation is intended for environments where `os.fork()` is not available.
"""
def fork_work_horse(self, job: 'Job', queue: 'Queue'):
"""Spawns a work horse to perform the actual work using os.spawn()."""
os.environ['RQ_WORKER_ID'] = self.name
os.environ['RQ_JOB_ID'] = job.id
os.environ['RQ_EXECUTION_ID'] = self.execution.id # type: ignore
redis_kwargs = self.connection.connection_pool.connection_kwargs
if redis_kwargs.get('retry'):
# Remove retry from connection kwargs to avoid issues with os.spawnv
del redis_kwargs['retry']
child_pid = os.spawnv(
os.P_NOWAIT,
sys.executable,
[
sys.executable,
'-c',
f"""
import os
import sys
from redis import Redis
from rq import Worker, Queue
from rq.job import Job
from rq.executions import Execution
# Recreate worker instance
redis = Redis(**{redis_kwargs})
worker = Worker.find_by_key("{self.key}", connection=redis)
if not worker:
sys.exit(1)
# Reconstruct job, queue and execution objects
job = Job.fetch("{job.id}", connection=worker.connection)
queue = Queue("{queue.name}", connection=worker.connection)
execution_id = os.environ.get('RQ_EXECUTION_ID')
worker.execution = Execution.fetch(execution_id, job.id, connection=worker.connection)
# Set up work horse
os.setpgrp()
worker._is_horse = True
worker.main_work_horse(job, queue)
""",
],
)
self._horse_pid = child_pid
self.procline(f'Spawned {child_pid} at {time.time()}')
class SimpleWorker(Worker):
def execute_job(self, job: 'Job', queue: 'Queue'):
"""Execute job in same thread/process, do not fork()"""
self.prepare_execution(job)
self.perform_job(job, queue)
self.set_state(WorkerStatus.IDLE)
def get_heartbeat_ttl(self, job: 'Job') -> int:
"""-1" means that jobs never timeout. In this case, we should _not_ do -1 + 60 = 59.
We should just stick to DEFAULT_WORKER_TTL.
Args:
job (Job): The Job
Returns:
ttl (int): TTL
"""
if job.timeout == -1:
return DEFAULT_WORKER_TTL
else:
return int(job.timeout or DEFAULT_WORKER_TTL) + 60
class HerokuWorker(Worker):
"""
Modified version of rq worker which:
* stops work horses getting killed with SIGTERM
* sends SIGRTMIN to work horses on SIGTERM to the main process which in turn
causes the horse to crash `imminent_shutdown_delay` seconds later
"""
imminent_shutdown_delay = 6
frame_properties = ['f_code', 'f_lasti', 'f_lineno', 'f_locals', 'f_trace']
def setup_work_horse_signals(self):
"""Modified to ignore SIGINT and SIGTERM and only handle SIGRTMIN"""
signal.signal(signal.SIGRTMIN, self.request_stop_sigrtmin)
signal.signal(signal.SIGINT, signal.SIG_IGN)
signal.signal(signal.SIGTERM, signal.SIG_IGN)
def handle_warm_shutdown_request(self):
"""If horse is alive send it SIGRTMIN"""
if self.horse_pid != 0:
self.log.info('Worker %s: warm shut down requested, sending horse SIGRTMIN signal', self.key)
self.kill_horse(sig=signal.SIGRTMIN)
else:
self.log.warning('Warm shut down requested, no horse found')
def request_stop_sigrtmin(self, signum, frame):
if self.imminent_shutdown_delay == 0:
self.log.warning('Imminent shutdown, raising ShutDownImminentException immediately')
self.request_force_stop_sigrtmin(signum, frame)
else:
self.log.warning(
'Imminent shutdown, raising ShutDownImminentException in %d seconds', self.imminent_shutdown_delay
)
signal.signal(signal.SIGRTMIN, self.request_force_stop_sigrtmin)
signal.signal(signal.SIGALRM, self.request_force_stop_sigrtmin)
signal.alarm(self.imminent_shutdown_delay)
def request_force_stop_sigrtmin(self, signum, frame):
info = dict((attr, getattr(frame, attr)) for attr in self.frame_properties)
self.log.warning('raising ShutDownImminentException to cancel job...')
raise ShutDownImminentException('shut down imminent (signal: %s)' % signal_name(signum), info)
class RoundRobinWorker(Worker):
"""
Modified version of Worker that dequeues jobs from the queues using a round-robin strategy.
"""
def reorder_queues(self, reference_queue):
pos = self._ordered_queues.index(reference_queue)
self._ordered_queues = self._ordered_queues[pos + 1 :] + self._ordered_queues[: pos + 1]
class RandomWorker(Worker):
"""
Modified version of Worker that dequeues jobs from the queues using a random strategy.
"""
def reorder_queues(self, reference_queue):
shuffle(self._ordered_queues)
|