How do I SSH into my installation to use mysqldump?

[Sean McGown]

I read the documentation here:

https://confluence.atlassian.com/adminjiraserver/backing-up-data-938847673.html

..but didn't see any links to how I can SSH into the installation and run mysqldump for a full backup. We want a more complete backup than the XML solution provides and also are interested in setting up cron to run a regular backup and copy it to a remote location. I searched "mysqldump" and "ssh key" but didn't get anything specific about what to point SSH at, what U+P I should use...

Any guidance from someone who has done what I'm describing? Or at least someone who has SSH'd into their installation?

[Bastian Stehmann · Accepted solution]

Hi @Sean McGown ,

If you are on cloud, there is no way to access the site per SSH and there is also no way to access the database directly to dump it. In cloud, you can only use the UI methods to create a backup. 

[Sean McGown]

And if I read correctly, there's no way to do a complete mysqldump with the UI tools, correct?

[Bastian Stehmann]

Yes, correct, sadly not.

[Nic Brough -Adaptavist-]

Even if you had access, mysqldump wouldn't work - they're not using mysql.

The XML backup is everything you need to create a new Jira system, and has all your data.  What more do you think you would get from a database dump?

[Craig Castle-Mead]

As Nic mentioned, you'll get all available data from the XML backup.

From here, you could restore the XML backup to a Jira data center trial node of your own using MySQL as the database and then mysqldump, but you're just going to get the same data as the XML export in a different format.

What are you trying to achieve?

CCM

[Sean McGown]

@Craig Castle-Mead 

Looking at the documentation, it recommends using tools like mysqldump for backup over the XML backup because they are more complete (whatever that means). See here. Was trying to affect a mysqldump backup but I guess on the cloud platform, it's not possible. Or is it?

The documentation says it won't be complete and you and @Nic Brough -Adaptavist-  are saying it will be. I'm new here so I'm starting to get concerned about what the correct recommended backup method is.

[Nic Brough -Adaptavist-]

No, you've misunderstood the documentation there.

A database backup is no more "complete" than the xml backup.  What they are saying in that doc is that a database backup is better for a large active Jira because database tools can ensure point-in-time backups which ignore changes to data while the backup is running

The xml backups are just as complete, but allow for errors because they can not check for changes during the process - think of the xml as reading out each table in turn, from A-Z.  If someone edits an issue when the xml has done tables A-F, then the backup will go wrong, because it has finished (say) customfieldvalue, without the edits, but will be exporting jiraissue with edits later.

Also, note that that document is for Server and DC, not Cloud.  Cloud backups are also coded differently to the Server/DC XML export, and don't work that way.

It's not about "recommended" - it's the only (full) backup you can do on Cloud.

[Sean McGown]

Thanks so much for the clarification. That helps a lot.