If using systemd, it does not honor the default timeout in the start script of 10 minutes. The default timeout for systemd services is 90 seconds.
Also on Fedora 23+ valgrind doesn't interact well with systemd anymore.
Need to add these lines to /etc/sysconfig/dirsrv.systemd:
TimeoutStartSec=10m -> set timeout to what we use in our start script NotifyAccess=all -> This allows systemd to receive/accept startup/shutdown notifications when using valgrind
attachment 0001-Ticket-48935-Update-dirsrv.systemd-file.patch
I just would like to learn how the new value 10 min works. I remember we fixed a ticket to shorten the timeout in the startup script which prevented to quit when, e.g., some selinux error occurred. If this change does not bring back the problem, you have my ack. :)
Replying to [comment:2 nhosoi]:
I just would like to learn how the new value 10 min works.
This is just sets the timeout when "starting" a service using systemctl
I remember we fixed a ticket to shorten the timeout in the startup script which prevented to quit when, e.g., some selinux error occurred.
We found this issue when valgrind was misbehaving. While the real issue had to do with another systemd setting(NotifyAccess), I thought we should make the timeouts consistent since we were updating the systemd file.
Now there is a timeout of 10 minutes in the start-dirsrv script when waiting for a PID file to be written, but perhaps this is obsolete(including the 10 minute timeout). If I'm correct the pid file is written before we return success/failure.
If this timeout was erroneous to begin with, then what is a good timeout? How long do we wait if a database is being recovered? 90 seconds seems too short in my opinion.
If this change does not bring back the problem, you have my ack. :)
Do you have the steps to reproduce the selinux issue?
I think this ticket and the fix mentioned in this ticket was in my mind.
https://fedorahosted.org/389/ticket/48336 Summary: setup-ds should detect if port is already defined
The retry count has been reduced in a recent fix, but retrying makes no sense if it is known to fail again.
Replying to [comment:4 nhosoi]:
I think this ticket and the fix mentioned in this ticket was in my mind. https://fedorahosted.org/389/ticket/48336 Summary: setup-ds should detect if port is already defined The retry count has been reduced in a recent fix, but retrying makes no sense if it is known to fail again.
Ahh, this is completely different. This issue (ticket 48336) happens internal to DS - it has nothing to do with the systemctl timeout. I also just verified there is no regression with 48336.
Anyway, I leave my patch as it stands (with the 10min timeout).
Thank you for your confirmation! As I wrote in my previous comment, you have my ack. :)
Thanks Noriko!
1972195..ce44176 master -> master commit ce44176 Author: Mark Reynolds mreynolds@redhat.com Date: Tue Jul 26 15:31:32 2016 -0400
This also impacts 1.3.4
c8e7fc5..5e810f8 389-ds-base-1.3.4 -> 389-ds-base-1.3.4 commit 5e810f8
Metadata Update from @nhosoi: - Issue assigned to mreynolds - Issue set to the milestone: 0.0 NEEDS_TRIAGE
Metadata Update from @vashirov: - Issue set to the milestone: None (was: 0.0 NEEDS_TRIAGE)
389-ds-base is moving from Pagure to Github. This means that new issues and pull requests will be accepted only in 389-ds-base's github repository.
This issue has been cloned to Github and is available here: - https://github.com/389ds/389-ds-base/issues/1994
If you want to receive further updates on the issue, please navigate to the github issue and click on subscribe button.
subscribe
Thank you for understanding. We apologize for all inconvenience.
Metadata Update from @spichugi: - Issue close_status updated to: wontfix (was: Fixed)
Login to comment on this ticket.