Bug 217297
| Summary: | [www/fcgiwrap] Stale fcgiwrap processes left even after service is stopped | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | spry | ||||
| Component: | Individual Port(s) | Assignee: | Adam Weinberger <adamw> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Only Me | CC: | lstewart | ||||
| Priority: | --- | Flags: | bugzilla:
maintainer-feedback?
(adamw) |
||||
| Version: | Latest | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Attachments: |
|
||||||
I can confirm this behaviour when fcgiwrap_flags="-c 4". The main process gets killed but the extras don't. I'm hesitant to use `killall fcgiwrap` because it prevents killing just one profile. I'm not sure how else to do it, though. Yeah you're right, I only have 1 profile at the moment for this tho, multiple profiles need to also be considered. Arrr... Created attachment 182791 [details]
Use a custom stop command that kills the pgrp
Stopping fcgiwrap is very flaky for me on 12-CURRENT out of the box, and almost always gets stuck "Waiting for PIDS" until rc.subr gives up.
The attached patch against the fcgiwrap rc.d script is working well for me and should be profile friendly (though has not been tested with multiple profiles).
(In reply to Lawrence Stewart from comment #3) Thanks. Works for me! Not tested w/ > 1 profiles tho. A commit references this bug: Author: adamw Date: Fri May 26 16:54:40 UTC 2017 New revision: 441799 URL: https://svnweb.freebsd.org/changeset/ports/441799 Log: Fix stale processes not being terminated by the rc(8) script. Previously, when fcgiwrap_flags="-c [n]" with n > 1, only the controlling process was terminated, with the children running. This patch fixes this behaviour, terminating the children. PR: 217297 Submitted by: lstewart Reported by: spry anarchy in the ph MFH: 2017Q2 Changes: head/www/fcgiwrap/Makefile head/www/fcgiwrap/files/fcgiwrap.in Sorry about the delay in committing this! Thanks for the patch, Lawrence. It's committed now. I'm leaving the PR open pending a merge to quarterly. A commit references this bug: Author: adamw Date: Sat May 27 16:05:47 UTC 2017 New revision: 441851 URL: https://svnweb.freebsd.org/changeset/ports/441851 Log: MFH: r438923 r441799 r438923: Release these ports to the collective. r441799: Fix stale processes not being terminated by the rc(8) script. Previously, when fcgiwrap_flags="-c [n]" with n > 1, only the controlling process was terminated, with the children running. This patch fixes this behaviour, terminating the children. PR: 217297 Submitted by: lstewart Reported by: spry anarchy in the ph Approved by: ports-secteam (miwi) Changes: _U branches/2017Q2/ branches/2017Q2/sysutils/vimpager/Makefile branches/2017Q2/www/fcgi/Makefile branches/2017Q2/www/fcgiwrap/Makefile branches/2017Q2/www/fcgiwrap/files/fcgiwrap.in MFH is done. Thanks again for the patch, Lawrence. Thanks for taking care of it Adam. |
Hi Seems there is a problem w/ stale processes not being killed USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND www 18674 0.0 0.0 14492 2036 - SsJ 8:49AM 0:00.00 daemon: /usr/local/sbin/fcgiwrap[18675] (daemon) www 18675 0.0 0.0 14472 2012 - SJ 8:49AM 0:00.00 - /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 www 18677 0.0 0.0 14472 2016 - SJ 8:49AM 0:00.00 |-- /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 www 18678 0.0 0.0 14472 2016 - SJ 8:49AM 0:00.00 |-- /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 www 18679 0.0 0.0 14472 2016 - SJ 8:49AM 0:00.00 |-- /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 www 18680 0.0 0.0 14472 2016 - SJ 8:49AM 0:00.00 `-- /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 root@:~# /usr/local/etc/rc.d/fcgiwrap stop ===> fcgiwrap profile: xxxx Stopping fcgiwrap. ... www 61512 0.0 0.0 14472 2060 - IJ Thu09AM 0:00.00 /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 www 61513 0.0 0.0 14472 2016 - IJ Thu09AM 0:00.00 /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 www 61514 0.0 0.0 14472 2016 - IJ Thu09AM 0:00.00 /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 www 61515 0.0 0.0 14472 2016 - IJ Thu09AM 0:00.00 /usr/local/sbin/fcgiwrap -s unix:/var/run/fcgiwrap/fcgiwrap.alerts.socket -c 4 ... Should there be something like ? --- fcgiwrap 2017-01-12 23:12:54.000000000 -0500 +++ /usr/local/etc/rc.d/fcgiwrap 2017-02-22 09:06:05.940283000 -0500 @@ -92,6 +92,7 @@ case ${fcgiwrap_socket} in unix*) test -S ${fcgiwrap_socket#unix:} && rm -f ${fcgiwrap_socket#unix:} + killall fcgiwrap ;; esac rm -f $pidfile