Synchronise 2 instances of JIRA

[Saneth Kumar]

I'm currently using JIRA 4.3 and I need to set up a Fall-Back instance of JIRA incase my Live (primary) JIRA instance goes down.

These 2 instances of JIRA need to always be in sync. If my Live Server goes down, users should be able to use JIRA on the fall back instance within a few minutes, and once the Live instance is back up all the issues created on the fall-back instance should be replicated in the Live instance.

Would using a 3rd party file synchronization tool (such as Allway Sync or GoodSync) be able to accomplish this by sychronizing the JIRA Directories (except server.xml which contains the base URL).

[Aspect Infra Team · Accepted solution]

you can setup DB and filesystem sync between servers on OS level

this is enough for having DR ready to start with reasonable data lost ( 10-15 min ) and in a short time ( less than a 5 min - just a JIRA startup time ). on linux we use cron to run sync script every 10 minutes.

rollback are manual but with absolutely same script.

[Saneth Kumar]

Hi,

Thank you for your response.

Would this work if my JIRA instances were on 2 different servers? Which folders would I need to sychronize?

I'm guessing we need to sync only the Home Directory, becausing syncing the install directory would lead to discrepencies in the Base URL (server.xml). Please correct me if I'm wrong.

[Łukasz Krupa]

You can accomplish that with different strategies.

- Cold standby: DB replication and cold standby application server (you need right procedures when deploying changes to PROD, so they will be reflected on backup env.)

- Jira Data center edition is professional solution to that problem. However, you may still consider DB replication, not mentioning file system backup in order to be prepared for the worst.

- Last solution assumes you got two independent environments (application servers, DBs, file system). You make a clone (i.e. xml backup restore) on one of them and run synchronization tool like IssueSYNC to exchange data between instances. When one instance fails you switch to second one. All changes are queued then until another Jira is up & running. 

From those three I believe the first option is what people use to implement in their organizations.

At some scale, more and more popular becomes second option (Jira Data center).

Last option is better for scenarios where you need only a part of data replicated, like some projects exposed to customers on separated Jira instance and in the same time your team works with other projects on 'primary' Jira. If exposed instance fails you recovers or gives access to your 'primary'.

[tousifs]

Hi Seneth,

See https://answers.atlassian.com/questions/15672/connecting-two-jira-instances

[Mizan]

Hi Saneth ,

A similar question here .

[Saneth Kumar]

Thanks for the link. However, I've not really used groovy scripts. Is there an easier method. Also, will the JEMH be able to create new projects?

[Saneth Kumar]

I need to configure a DR server for JIRA, any changes made to the primary server should reflect in the Backup Server (including new Projects, users etc). Does JEMH do all this