Bug #12859
closedkeepstore no longer functioning with repeated errors: http: Accept error: accept tcp [::]:25107: accept4: too many open files; retrying in 1s
Description
Every entry in the keepstore logs on one of my keepstores looks something like this:
2017-12-20_19:08:26.79910 2017/12/20 19:08:26 http: Accept error: accept tcp [::]:25107: accept4: too many open files; retrying in 1s
The keepstore is not accepting connections.
After a restart, the keepstore started functioning again.
Version:
ii keepstore 0.1.20171102145827.cc6f86f-1 amd64 Keep storage daemon, accessible to clients on the LAN
Updated by Ward Vandewege almost 7 years ago
- Status changed from New to In Progress
- Assigned To set to Ward Vandewege
- Target version set to 2017-12-20 Sprint
Hey Josh,
We see this sometimes on heavily loaded keepstores.
On Debian(ish) systems, the default limit for file descriptors is 1024.
If you have a recent kernel and OS, you can use the `prlimit` tool to modify the file descriptor limit upwards for a running process.
Alternatively, ulimit can do it too at process start time.
Does this help?
Thanks,
Ward.
Updated by Joshua Randall almost 7 years ago
I suspect so, my fix was also to add `ulimit -n 1048576` to the runit script that starts keepstore, and the problem hasn't come back so far.
I reported it as a bug because (a) this is not documented in the install docs and (b) it seems like there is a problem with the way keepstore handles open files if it can get into a state where it fails to accept every single incoming connection for many days. If it just occasionally had a "too many open files" issue when under heavy load, but then later it started working again (presumably because it closed some sockets) then (b) would not be an issue. but it would still be good to have a doc fix for (a). As it is, it seems like something must be wrong with the way keepstore handles closing files in order for it to be possible to get into a state where it is perpetually out of file descriptors?
Updated by Tom Morris almost 7 years ago
- Assigned To deleted (
Ward Vandewege) - Target version changed from 2017-12-20 Sprint to 2018-01-17 Sprint
Updated by Tom Morris almost 7 years ago
- Target version changed from 2018-01-17 Sprint to Arvados Future Sprints
Updated by Ward Vandewege almost 7 years ago
We suspect there is a bug here in an error path. We'll investigate.
In any case, perhaps we should increase the default value in the systemd config file at /lib/systemd/system/keepstore.service.
Perhaps 8192 is a better default:
# Copyright (C) The Arvados Authors. All rights reserved. # # SPDX-License-Identifier: AGPL-3.0 [Unit] Description=Arvados Keep Storage Daemon Documentation=https://doc.arvados.org/ After=network.target AssertPathExists=/etc/arvados/keepstore/keepstore.yml # systemd<230 StartLimitInterval=0 # systemd>=230 StartLimitIntervalSec=0 [Service] Type=notify LimitNOFILE=8192 ExecStart=/usr/bin/keepstore Restart=always RestartSec=1 [Install] WantedBy=multi-user.target
Overriding per system is also possible, one can put a file at /etc/systemd/system/keepstore.service.d/override.conf with these contents, for instance:
[Service] LimitNOFILE=8192
Updated by Peter Amstutz over 3 years ago
- Target version deleted (
Arvados Future Sprints)
Updated by Tom Clegg about 3 years ago
12859-keepstore-fd-leak @ 9ff58642c2c1eaae8d6dd7c01e8fb85a030c2dfc -- developer-run-tests: #2762
Updated by Tom Clegg about 3 years ago
- Target version set to 2021-11-10 sprint
- Assigned To set to Tom Clegg
Updated by Ward Vandewege about 3 years ago
Tom Clegg wrote:
12859-keepstore-fd-leak @ 9ff58642c2c1eaae8d6dd7c01e8fb85a030c2dfc -- developer-run-tests: #2762
Only one question, on line 295 in `services/keepstore/unix_volume.go`, is there a reason to keep that explicit call to tmpfile.Close() ? We now have a new `defer tmpfile.Close()` higher up, so...
LGTM, otherwise!
Updated by Tom Clegg about 3 years ago
Ward Vandewege wrote:
is there a reason to keep that explicit call to tmpfile.Close() ? We now have a new `defer tmpfile.Close()` higher up, so...
Only to detect errors (unflushed write?). The deferred Close just covers the return-early-because-we-hit-some-other-error-first case, where we don't care whether Close fails.
Updated by Tom Clegg about 3 years ago
As noted in the commit message, 9ff58642c fixes a file descriptor leak when a filesystem-backed volume has Serialize: true
and the client hangs up while waiting for another operation to finish on the same volume. When clients are piling up faster than keepstore can serve requests, many of the clients would time out at that stage, so this seems like a plausible explanation for the reported problem.
Updated by Tom Clegg about 3 years ago
- % Done changed from 0 to 100
- Status changed from In Progress to Resolved
Applied in changeset arvados-private:commit:arvados|13dea78ff8ad72bd8a406843bb702aef0d0f569a.