I've got MySQL running on DRBD in a two node cluster. I put the contents of /var/lib/mysql, among other things, on a DRBD disk and make a soft link on all my nodes at /var/lib/mysql pointing to the dir on DRBD. I use /cluster/mysql as the DRBD mount point and /cluster/mysql/var/lib/mysql for the actual DB files. It's my strange way of keeping things organized on each individual DRBD volumes that I have, one for each "service".
I start my IP, DRBD volume (with drbddisk), file system and mysql in that order with a group. There has been a lot of debate over this vs. MySQL replication from what I can tell. I haven't spent a lot of time weighing any pros and cons, this was simpler (I was building cluster services at the time, not reading MySQL documentation) and it's always worked from the start without fussing, failover after failover (mostly testing). I could pull out numbers on our MySQL work load (fairly low) but I don't have them accessible right now. Performance is fine for our purposes which is a back end for the "Zabbix" monitoring suite. I think many concerns about MySQL on DRBD are performance related, which I can't comment on. --Ryan > > Mysql cluster requires three nodes. It also, from my understanding, > requires significantly more memory per node than I have in budget. > > How are most people implementing mysql failover in heartbeat? > > > On 10/28/08 12:51 AM, "Nick" <[EMAIL PROTECTED]> wrote: > > I think that if you're starting from new DBs implementing the mysql > cluster is probably easier.. for me DRBD's strength is in replicating > things which don't already have builtin solutions (postfix's mail spool > for example).. > >> >> On 10/27/08 4:46 PM, "Wolf Noble" <[EMAIL PROTECTED]> wrote: >> >> I'm trying to set up a relatively simple cluster with 4 virtual nodes >> residing on 2 physical boxes. >> >> 2 nodes will be running mysql, with a vip on the active node. >> 2 nodes will be running drbd, a vip, and an apache instance. >> >> this is on centos 5.2 in testing, will be on RHEL in prod. >> >> heartbeat.i386 2.1.3-3.el5.centos >> installed >> heartbeat-gui.i386 2.1.3-3.el5.centos >> installed >> heartbeat-pils.i386 2.1.3-3.el5.centos >> installed >> heartbeat-stonith.i386 2.1.3-3.el5.centos >> installed >> >> I'm having troubles getting the apache ocf RA to work, and from what >> I've read, many other people have as well, but I've not read a workable >> solution, so I've defaulted to just using the linux init script as a >> resource. >> >> if anyone would mind pointing me to a setup of the vmware stonith >> device, I'd really appreciate it... I'm having troubles making heads or >> tails of it. >> >> also if anyone has a pointer on the apache RA, I'd love to hear about >> it... >> >> Thanks so much guys. >> >> -Wolf Noble > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems > _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
