2020-04-02 09:59:28 -04:00
|
|
|
#!/usr/bin/env python3
|
2012-11-26 10:55:22 -05:00
|
|
|
|
2012-10-02 14:09:24 -04:00
|
|
|
import os
|
zmirror: Allow duplicate zmirror processes to die gracefully.
Fixes #602.
I replaced the SIGKILL with a SIGINT, and then catch SIGINT with a
handler. This handler calls cancelSubs if necessary, and can later be
edited to perform other clean-up operations, too. I thought about, in
this same commit, changing the SIGTERM in
maybe_restart_mirroring_script to a SIGINT, but after tracing out the
code paths, I realized that isn't necessary. (The SIGTERM is
necessarily performed on a process that has not subscribed to any
zephyr classes, so cancelSubs is unnecessary. If we do think that we
may want to add additional clean-up operations in the future, though,
then it might be worth investigating changing this SIGTERM.)
(imported from commit 692b295be6cb40b0e4ec2ca0bc58c58056ed9bd9)
2013-01-03 13:21:19 -05:00
|
|
|
import signal
|
2021-05-28 05:00:04 -04:00
|
|
|
import subprocess
|
|
|
|
import sys
|
|
|
|
import traceback
|
2012-10-29 18:09:51 -04:00
|
|
|
|
2017-03-20 15:54:31 -04:00
|
|
|
sys.path[:0] = [os.path.dirname(__file__)]
|
2017-05-10 14:52:11 -04:00
|
|
|
from zephyr_mirror_backend import parse_args
|
2012-10-11 16:20:38 -04:00
|
|
|
|
2016-07-29 23:14:08 -04:00
|
|
|
(options, args) = parse_args()
|
|
|
|
|
2016-07-29 23:13:08 -04:00
|
|
|
from types import FrameType
|
2016-07-29 23:14:08 -04:00
|
|
|
|
2021-05-28 05:00:04 -04:00
|
|
|
|
2020-04-18 18:59:12 -04:00
|
|
|
def die(signal: int, frame: FrameType) -> None:
|
2016-07-29 23:13:08 -04:00
|
|
|
|
zmirror: Allow duplicate zmirror processes to die gracefully.
Fixes #602.
I replaced the SIGKILL with a SIGINT, and then catch SIGINT with a
handler. This handler calls cancelSubs if necessary, and can later be
edited to perform other clean-up operations, too. I thought about, in
this same commit, changing the SIGTERM in
maybe_restart_mirroring_script to a SIGINT, but after tracing out the
code paths, I realized that isn't necessary. (The SIGTERM is
necessarily performed on a process that has not subscribed to any
zephyr classes, so cancelSubs is unnecessary. If we do think that we
may want to add additional clean-up operations in the future, though,
then it might be worth investigating changing this SIGTERM.)
(imported from commit 692b295be6cb40b0e4ec2ca0bc58c58056ed9bd9)
2013-01-03 13:21:19 -05:00
|
|
|
# We actually want to exit, so run os._exit (so as not to be caught and restarted)
|
|
|
|
os._exit(1)
|
|
|
|
|
2021-05-28 05:03:46 -04:00
|
|
|
|
zmirror: Allow duplicate zmirror processes to die gracefully.
Fixes #602.
I replaced the SIGKILL with a SIGINT, and then catch SIGINT with a
handler. This handler calls cancelSubs if necessary, and can later be
edited to perform other clean-up operations, too. I thought about, in
this same commit, changing the SIGTERM in
maybe_restart_mirroring_script to a SIGINT, but after tracing out the
code paths, I realized that isn't necessary. (The SIGTERM is
necessarily performed on a process that has not subscribed to any
zephyr classes, so cancelSubs is unnecessary. If we do think that we
may want to add additional clean-up operations in the future, though,
then it might be worth investigating changing this SIGTERM.)
(imported from commit 692b295be6cb40b0e4ec2ca0bc58c58056ed9bd9)
2013-01-03 13:21:19 -05:00
|
|
|
signal.signal(signal.SIGINT, die)
|
|
|
|
|
2014-03-11 14:09:15 -04:00
|
|
|
from zulip import RandomExponentialBackoff
|
|
|
|
|
2017-10-05 16:16:04 -04:00
|
|
|
args = [os.path.join(os.path.dirname(os.path.realpath(__file__)), "zephyr_mirror_backend.py")]
|
2012-11-20 15:23:30 -05:00
|
|
|
args.extend(sys.argv[1:])
|
|
|
|
|
|
|
|
if options.sync_subscriptions:
|
|
|
|
subprocess.call(args)
|
|
|
|
sys.exit(0)
|
|
|
|
|
2012-11-29 08:58:59 -05:00
|
|
|
if options.forward_class_messages and not options.noshard:
|
2017-10-05 16:23:17 -04:00
|
|
|
# Needed to get access to zephyr.lib.parallel
|
2013-10-04 13:19:57 -04:00
|
|
|
sys.path.append("/home/zulip/zulip")
|
2013-08-20 14:03:07 -04:00
|
|
|
if options.on_startup_command is not None:
|
|
|
|
subprocess.call([options.on_startup_command])
|
2013-08-20 06:50:24 -04:00
|
|
|
from zerver.lib.parallel import run_parallel
|
2021-05-28 05:03:46 -04:00
|
|
|
|
2015-11-01 11:11:06 -05:00
|
|
|
print("Starting parallel zephyr class mirroring bot")
|
2012-11-27 10:27:02 -05:00
|
|
|
jobs = list("0123456789abcdef")
|
2016-11-29 01:22:02 -05:00
|
|
|
|
2020-04-18 18:59:12 -04:00
|
|
|
def run_job(shard: str) -> int:
|
2021-05-28 07:19:40 -04:00
|
|
|
subprocess.call(args + [f"--shard={shard}"])
|
2012-11-27 10:27:02 -05:00
|
|
|
return 0
|
2021-05-28 05:03:46 -04:00
|
|
|
|
2012-11-27 10:27:02 -05:00
|
|
|
for (status, job) in run_parallel(run_job, jobs, threads=16):
|
2015-11-01 11:11:06 -05:00
|
|
|
print("A mirroring shard died!")
|
2012-11-27 10:27:02 -05:00
|
|
|
sys.exit(0)
|
|
|
|
|
2014-03-17 13:31:00 -04:00
|
|
|
backoff = RandomExponentialBackoff(timeout_success_equivalent=300)
|
2013-02-14 12:04:45 -05:00
|
|
|
while backoff.keep_going():
|
2015-11-01 11:11:06 -05:00
|
|
|
print("Starting zephyr mirroring bot")
|
2012-11-13 16:30:39 -05:00
|
|
|
try:
|
2012-11-19 12:04:23 -05:00
|
|
|
subprocess.call(args)
|
2017-01-08 10:52:26 -05:00
|
|
|
except Exception:
|
2012-11-19 12:04:23 -05:00
|
|
|
traceback.print_exc()
|
2013-02-14 12:04:45 -05:00
|
|
|
backoff.fail()
|
|
|
|
|
2017-01-08 10:52:26 -05:00
|
|
|
|
|
|
|
error_message = """
|
|
|
|
ERROR: The Zephyr mirroring bot is unable to continue mirroring Zephyrs.
|
|
|
|
This is often caused by failing to maintain unexpired Kerberos tickets
|
2020-06-08 17:03:27 -04:00
|
|
|
or AFS tokens. See https://zulip.com/zephyr for documentation on how to
|
2017-01-08 10:52:26 -05:00
|
|
|
maintain unexpired Kerberos tickets and AFS tokens.
|
|
|
|
"""
|
|
|
|
print(error_message)
|
2013-02-14 12:04:45 -05:00
|
|
|
sys.exit(1)
|