Incron causes defunct processes

[Gerry]

Description: Ubuntu Disco Dingo (development branch)
Release: 19.04
ii incron 0.5.12-1 amd64 cron-like daemon which handles filesystem events

gsker@veeta:~>incrontab -l
/home/gsker/tmp/incrontest/testfile.txt IN_CLOSE_WRITE /home/gsker/tmp/incrontest/incron.sh $# $%

gsker@veeta:~>ps -ef | grep incron
root 5558 1 0 14:47 ? 00:00:00 /usr/sbin/incrond
gsker 6366 5558 0 14:52 ? 00:00:00 [incron.sh] <defunct>
gsker 6371 6190 0 14:52 pts/4 00:00:00 tail -f incrontest.log
gsker 6386 5558 0 14:52 ? 00:00:00 [incron.sh] <defunct>
gsker 6390 5558 0 14:52 ? 00:00:00 [incron.sh] <defunct>

The incron works, but I end up with hundreds of defunct processes.

This bug report for RedHat implies a fix:
https://bugzilla.redhat.com/show_bug.cgi?id=1656939

I grabbed the incrond executable from the x86_64 rpm here:
  https://koji.fedoraproject.org/koji/buildinfo?buildID=1235814

It appears to be working correctly.

[Gerry (gsker)]

Okay I see the git repo that this descends from, https://github.com/ar-/incron, has this fix and several others. And I see that the upstream is the debian package. How do we go about getting the whole chain updated?

This bug was fixed back in Oct of 2017. https://github.com/ar-/incron/commit/196975d26fd04176a1c877fa3c404efd8103c9c2#diff-d73e309589e3f71b662dce37ccc688b6

[Gerry (gsker)]

Given how this package is not getting much love, I switched to using systemd's systemd.path functionality.
Very briefly:

root@veeta:/etc/systemd/system>cat gsker-temptrax.path
[Path]
PathChanged= /var/www/gsker/weather/temptrax.log
[Install]
WantedBy=default.target

root@veeta:/etc/systemd/system>cat gsker-temptrax.service
[Service]
Type=oneshot
User=gsker
ExecStart= /home/gsker/weather/temptrax/plotrun.sh

root@veeta:/etc/systemd/system>systemctl daemon-reload
root@veeta:/etc/systemd/system>systemctl enable gsker-temptrax.path
root@veeta:/etc/systemd/system>systemctl start gsker-temptrax.path

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Philippe Cloutier (chealer-gmail)]

Assuming this is about Debian #930526, this is reportedly fixed in 0.5.12-2, and therefore in all releases except bionic.

[Gerry (gsker)]

There are some serious bugs in incron.

It should probably be deprecated and removed from the repos for security reasons.

* https://github.com/ar-/incron/issues/67
* https://github.com/ar-/incron/issues/72
* https://github.com/ar-/incron/issues/68

[Philippe Cloutier (chealer-gmail)]

This is probably solved in all supported releases, as bionic uses 0.5.10-3build1, and 0.5.10 is apparently not affected.

That being said, I am experiencing a similar bug with bionic in a Docker container, where incrond itself becomes a zombie process (and therefore stops working).

[Philippe Cloutier (chealer-gmail)]

This was reported upstream as https://github.com/ar-/incron/issues/52

[sles (slesru)]

This bug still exists in 20.04.
Why?

[Jason Perrin (jvperrin)]

I've been able to confirm this happening also on 22.04 (Jammy) with package version 0.5.12-2build1.

I noticed that https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=981124 was fixed recently (Jan 2023) with another patch in Debian (https://sources.debian.org/patches/incron/0.5.12-4/05_avoid_unterminated_forked_processes.patch/), so I think pulling in 0.5.12-4 from Debian might fix the issue here too, and would be a valuable thing to pull in for LTS releases too in order to prevent runaway zombie process issues.

Thanks!