Today I noticed a sudden slowness of disk-IO on my ClearOS7 server. Looking at the logs there has been a constant repetition of 2 commands every (exactly) 5 mins for the last 8 hours and counting :
servicewatch: sanity checking winbind , followed by
servicewatch: restarting winbind after 30 secs.
The system was not actively used when this started.
This system was originally set up with Samba Domain, which I had removed 2 weeks ago, trying to replace it by Samba. That substitution having failed I knew I had to rebuild my server from scratch but had been too busy doing that while I could do without Samba. The system has worked fine and not been updated during the last 2 weeks. This problem started quite suddenly.
I'm out of clues as to what this is all about. I will rebuild my server but this is just nagging me.
servicewatch: sanity checking winbind , followed by
servicewatch: restarting winbind after 30 secs.
The system was not actively used when this started.
This system was originally set up with Samba Domain, which I had removed 2 weeks ago, trying to replace it by Samba. That substitution having failed I knew I had to rebuild my server from scratch but had been too busy doing that while I could do without Samba. The system has worked fine and not been updated during the last 2 weeks. This problem started quite suddenly.
I'm out of clues as to what this is all about. I will rebuild my server but this is just nagging me.
Share this post:
Responses (4)
-
Accepted Answer
-
Accepted Answer
Thank you very much Nick, for the excellent explanations and for all the work you're doing for all of us.
Trying to be helpful, here are my 2 cents:
/code
yum reinstall app-samba
systemctl restart winbind.service
...
journalctl -xe
.. systemd[1]: Unit winbind.service entered failed state.
.. systemd[1]: winbind.service failed.
.. servicewatch[22102]: restarting winbind
.. polkitd[919]: Registered Authentication Agent for unix-process:24734:8819234 (system bus name :1.18811 [/usr/bin/pkttyagent --notify-fd 5 --fallback],
.. systemd[1]: winbind.service: main process exited, code=exited, status=1/FAILURE
.. systemd[1]: Failed to start Samba Winbind Daemon.
-- Subject: Unit winbind.service has failed
-- Defined-By: systemd
-- Support: http://lists.freedesktop.org/mailman/listinfo/systemd-devel
--
-- Unit winbind.service has failed.
--
-- The result is failed.
.. systemd[1]: Unit winbind.service entered failed state.
.. systemd[1]: winbind.service failed.
.. polkitd[919]: Unregistered Authentication Agent for unix-process:24734:8819234 (system bus name :1.18811, object path /org/freedesktop/PolicyKit1/Authe
/code
Hope this helps -
Accepted Answer
The sequences of events has come back to me. Up to and including ClearOS 7.6 we used to build samba ourselves and noticed a problem with winbind startup which we initially fixed with a kludge and then, near 7.7, with a better fix. At this point we noticed Centos were now building Samba with domain support - they never used to which is the reason why we were building it ourselves. They'd also bumped into the winbind issue and put fixes through for Centos 7.7 (samba 4.91). At this point we decided to revert to the upstream centos build.
I noticed that winbind was failing if it was starting before LDAP, but as we were using the upstream Samba we could not change how winbind started. Instead we pushed a fix to openldap to tell it to try to start before winbind, but it is not a "mandatory" thing, more a suggestion to systemd so it is it is not 100% reliable. To do a belt and braces fix we also added it to servicewatch as I mentioned before. -
Accepted Answer
Winbind is becoming increasingly important for samba and with the 4.9.1 version in ClearOS 7.7 I could not connect to a simple share without winbind running. For that reason we've made a couple of changes. If you install samba, (or perhaps app-samba) it automatically sets winbind to autostart. Beyond this there is a ClearOS program, servicewatch, which monitors a number of programs and automatically tries to start them if they stop. For 7.7, we decided to add winbind to this program (and I rewrote it a bit as well). It is only a mini-script and you can see it in /usr/clearos/apps/base/deploy/servicewatch. Every 5 minutes, it checks to see if the list of programs is set to start on boot and tries to start is if it is not running. This is what you are seeing. The easiest thing to do is to just disable winbind from auto-starting with a:
Be aware program updates may re-enable it. Alternatively you can edit servicewatch and remove winbind from the list, but again, this will get overwritten every time app-base is updated.systemctl disable winbind
Winbind should fail pretty quickly when it fails so it should not be causing you a resource problem.
Please login to post a reply
You will need to be logged in to be able to post a reply. Login using the form on the right or register an account if you are new here.
Register Here »