Machine crash and sql dump

[Stephane DEWITTE]

Hello all,

I have (or had...) 3 virtual machines, with Jira, Confluence and Bitbucket. The hypervisor crashed, and the machines are not working anymore (the os won't start).

I have SQL dump of the machines (mariadb), but no XML export.

I tried to reinstall instances, but when I try to attach to the databases, it complains about already existing data.

Can someone tell me the procedure to gain working instances with data ? When I looked, I only find informations about XML dump.

[Nic Brough {Adaptavist} · Accepted solution]

You're probably very close, but I'm not sure exactly what you've done with your "reinstall".  There is a minor trick to this in your case, and I'm not sure a lot of the docs really cover it properly.

There is an additional question too - do you have a backup of the installation's "home" directories?  That'll come in later...

So, for each application:

1. Create an empty database
2. Install it as though you were creating a new installation, i.e with the empty database, and no data from the "home" directory backup if you have one)
3. Get it to the point where it runs and you can log in with the initial admin user
4. Stop the system
5. Restore the database backup into the database
6. If you have a backup of the home directory, restore that into the new one, but preserve the config file that has the database connection in it (dbconfig.xml for JIRA, confluence.cfg.xml for Confluence, can't remember Bitbucket and check carefully before relying on my memory)
7. Restart the application
8. If you had the home directory, you might find it is now working fine, but I'd still recommend the next step
9. Do a full re-index of the whole thing

[Stephane DEWITTE]

Hello, and thanks for your help. It worked fine on Jira, now I'm testing it on Confluence and BitBucket.

[Nic Brough {Adaptavist}]

No problem.  I've been through it a lot on the current project - not crashes, but rapid restores of test systems, so I'm quite familiar with it for JIRA.  Confluence, I wasn't the lead on that, but did a few.  I'm sorry if it's not quite right for Bitbucket - my colleagues did all of those.  I think it's broadly correct for older versions (when it was Stash), but I'm getting rusty.