This is a read-only archive of the Mumble forums.

This website archives and makes accessible historical state. It receives no updates or corrections. It is provided only to keep the information accessible as-is, under their old address.

For up-to-date information please refer to the Mumble website and its linked documentation and other resources. For support please refer to one of our other community/support channels.

Jump to content

Mumble randomly loses group membership data on reboot


saq
 Share

Recommended Posts

Ubuntu 14.0.4.1 running Mumble-Server 1.2.8-1~ppa1~trusty1 from apt. Using mysql backend and ldap auth integration via ice and an ice python script.


Sometimes when the service gets restarted the contents of the murmur_group_members table in the db gets very odd. Something that is normally about 8-10 entries with different groups gets trimmed down to 4 entries and the group id's don't appear to be correct.

When this happens certain users (and not necessarily who is missing from the table) who are in say the admin group no longer have their admin powers. If I do an INSERT INTO on the DB side to try and grant admin powers to resolve the issue, the second the account logs in the manually added record gets removed from the table.

If someone who somehow still retains their admin abilities then goes and adds my account back to the admin group the murmur_group_members table gets fixed and is dramatically different than how it was moments ago. It has the expected entries for users and the appropriate group id's in all the right places.


Any ideas whats going on here?

Link to comment
Share on other sites

  • 2 weeks later...
  • Moderators
If I do an INSERT INTO on the DB side to try and grant admin powers to resolve the issue, the second the account logs in the manually added record gets removed from the table.

 

I think this is intended behaviour, or at least, not considered a bug. Basically Murmur uses MySQL/SQLite as "file storage" and updates it as it sees fit, but I don't think we make any guarantee that changes to the SQL database are reflected in the running Murmur process.


The other part about the reboots is probably where the issue lies, and I think is likely a bug. When you say reboot, do you mean the entire box? Are you doing it cleanly with shutdown(1)?

Full disclosure: I used to run a commercial Mumble host, and my opinions do not reflect the opinions of the Mumble project.

Avatar is stolen from here

Link to comment
Share on other sites

I was doing it cleanly with a shutdown. I just did it again and the problem reared its ugly head again. I even did service mumble stop

If this is not the correct way let me know, I'm a little new at mumble server administration. I'm the server admin and I mainly have had other people taking care of the mumble user management but this problem has been quite perplexing and I need to figure it out.

Link to comment
Share on other sites

I just found something and cleaned it up, and I think I know what it was related to.

I used to be on the native sqlite database, but then decided to move it to mysql for various reasons. I had issues getting the migration going and one of my more talented programmer buddies helped get things migrated over. There were a few migration issues and because of that I think it generated a new "admin" group for every single channel, which meant I had 32 different groups named admin and they were all empty. I made a new channel called SysAdmins and moved all the relevant people over to that, and then deleted all of the admin groups. Next time I get a chance to reboot I'll see if any group membership is lost.

Link to comment
Share on other sites

 Share

×
×
  • Create New...