JIRA DR Failover Questions (JIRA Data Centre?)

Question

Hi,

We have a single JIRA Enterprise instance and Postgres database in our PROD data centre. We then export database dumps, attachments and plugins to our DR site.

This method has served us well, but we need a faster failover time as our indexing takes 30 mins. Ideally we'd like active-active replication between data centres with DNS failover handled by our load balancers.

We currently run on physical servers and will be moving to VM's, so HA failover will be covered via ESX HA. We are more specifically interested in DR recovery than inter-site HA failover.

I've been looking at JIRA Data Centre, but it still seems to only support a Cold Standby for DR? Granted the indexes are replicated which should speed up the process of starting in DR. I'm not sure the price justifies the move.

Has anyone achieved an Active-Active DR solution with JIRA Data Centre or another combination of products?

Cheers!

Roger Rooney · Answer

Thanks for the replies.

I have Postgres binary replication working in test (we'll be using this for our database replication going forward).

I'm just wondering if you guys use the JIRA index snapshot export/import feature in your DR setup? I suspect I can stick with standard JIRA install but rsync my index snapshots to DR where we'd import them when starting DR. Do you guys do this?

Also, do you have heartbeat running for Postgres replication? Or do you set the slave to master mode manually when you start your DR site?

Ta,

Volodymyr Krupach @BrizoIT · Answer

Did you check PostgreSQL master -> slave configurations: https://www.postgresql.org/docs/9.2/static/warm-standby.html#SYNCHRONOUS-REPLICATION?

Jobin Kuruvilla [Adaptavist] · Answer

JIRA DC is Active-Active clustering. i.e. you have more than one node running in the cluster and the fail-over happens fairly quick. This is what will be used in most cases.

But DR is a separate cluster in a different data center. It is an Active-Passive solution as you mentioned. This is solving the real disaster scenarios and not the failover when a node goes down. I haven't come across an Active-Active DR solution yet for JIRA.