789035 - upgrade stage*.db.scl3 from MySQL-5.0

Status: RESOLVED FIXED

(Data & BI Services Team :: DB: MySQL, task)

Description

[Dustin J. Mitchell [:dustin] (he/him)]

These are RHEL5 systems running MySQL-5.0.  That's not compatible with the mysql2::server class, and they should be upgraded anyway.

Comment 1

[Dustin J. Mitchell [:dustin] (he/him)]

Sheeri, what's the best way to do this, and what should I upgrade to?  Can the stage cluster run on a single node?  Is manual failover about the same as for a cluster with >2 servers?  Who do I need to talk to about a maintenance window on this cluster?

Comment 2

[Sheeri Cabral [:sheeri]]

Upgrade to Percona 5.1.

https://mana.mozilla.org/wiki/display/SYSADMIN/MySQL+5.0+to+5.1+Upgrade

The installing the Percona server section should probably be updated to "use puppet" - it's like steps 17-18, and there's still some manual stuff like "remove the old packages first.

Stage can run on a single node, so take stage2 out of the load balancer. Part of the instructions is to export with mysqldump and import, so you'll want to stop any slaves (e.g. if stage1 is a slave of stage2, we should disable that for the duration of the upgrade).

Once stage2 is upgraded and put back into the load balancer, we should wait a few days or a week so that any issues can be uncovered. To upgrade stage1, take it out of the load balancer, stop mysql and use xtrabackup to hot copy stage2 to it. I think that's covered on the page I linked to, actually.

Comment 3

[Sheeri Cabral [:sheeri]]

I'm working on stage2 right now, due to buildbot staging being on the machine (supposedly) and trying to get scalebase up in front of it, but I want to make sure the machine is standardized before I do that.

Comment 4

[Sheeri Cabral [:sheeri]]

stage2 is upgraded and in puppet now.

Comment 5

[Sheeri Cabral [:sheeri]]

Got good checksums, so I will be upgrading stage1 and copying the data there now.

Comment 6

[Sheeri Cabral [:sheeri]]

stage1 is upgraded, puppetized and has the backup!