Hi, Thank you for you answer, I have the same choice you had. And I think I'm not going to put fail-over on the mysql database, but just do a simple copy of the database to be sure not to loose all data if something append in the file system.
cheers, Antoine On 24/08/2013 08:26, Marcin Stolarek wrote: > Re: [slurm-dev] Managing the SLURM database > > > > 2013/8/22 Chauvin Antoine <[email protected] > <mailto:[email protected]>> > > > > Hi > > Hi, > > , > As recommended by SLURM I use a mysql database with the slurmdbd > daemon. > Slurm include the failover system, but nor mysql, I would like to know > how do you manage the mysql failover ? > > I was dealing with this problem a few weeks ago and the main > inferences are: > -) Consider if you really need mysql failover. Short outage of slurdbd > is not a problem, because slurmctld will store all data in memory and > send it to slurmdbd when it's back operating. > -) You cannot use ndb, since slurmdbd/mysql uses a keys on BLOB values > (and maybe something more from the incompatibility list). > -) You can set up "classical" Linux HA, with heartbeat/corosync to > migrate IP between master/backup mysql servers and: > -> Configure one way replication of mysql, and change > master/backup roles on failure > -> Use shared storage for master/slave mysql servers > database, and start backup on master mysql failure. > > cheers, > marcin -- ---------------------------------------------- Antoine CHAUVIN Groupe SR Synchrotron SOLEIL 01 69 35 93 23
