Confluence/Jira filesystem backup: Which directories are safe to exclude?

[Rich Remington]

I am trying to determine what directories are safe to exclude from a filesystem back of the HOME and INST folders of Confluence and Jira.

I've seen the answer from https://community.atlassian.com/t5/Confluence-questions/Confluence-home-directory-backup-What-directories-are-safe-to/qaq-p/434962, but this is quite dated now and does not address the INST folder.  I have 4.5GB under INST/temp and am guessing I don't need this to be backed up.

There does not seem to be a definitive answer to this, so I'm reaching out to support for the first time.  I have the same question about Jira as well.  Do I need to open a separate support request for that?

Thanks,
Rich

[Daniel Eads {unmonitored account}]

Hi Rich,

Thanks for searching first! That thread (despite its date) actually looks pretty close to me. For the installation folder, you might be able to get away with excluding the <install>/logs directory depending on your logging requirements.

In general though, I think you will want to keep the whole install directory (minus logs) to get you back and running quickly if something goes down. If you need to use your backups, you might not want to wait additional time after the restore is complete for your indexes to rebuild from scratch.

Have you found that you are significantly constrained with your backup capabilities? A lot of stuff in the install directory doesn't change, so a differential/incremental backup should be pretty small on the install dir.

[Rich Remington]

Thanks for the answer, Daniel, I appreciate it.

Just for some context.  I run a VERY small shop and don't perform this task regularly.  There isn't a ton of data in either system, so once a month backup has worked well for me and I haven't had any crashes where a restore was required.  I was hoping to finalize the automation of this, but noticed the size the backup was way out of proportion to expections.

I did do quite a bit of poking around before asking this question and it seems there are several folks with simlar, if not identical, questions and still no definitive answer. Unfortunately, "pretty close" is just not sufficient in my opinion.  Since the dated link above I referenced only talks about the HOME directory and does not cover other/new folders (e.g. analytics-logs, lock, recovery, webresource-temp, etc.) folders, it is incomplete for more recent versions.

I see that others have stated that removing the INST/temp files is safe if Confluence is not running at the time, so I'll give that a try as this is the biggest offender of wasted space, IMHO. This will help my situation, if the suggestion works, but I believe it's still not sufficient for the community at large.

I would strongly encourage Atlassian to work toward making a definitive statement about what is and what is not required to restore a complete backup for all of their products, especially for what I would consider their flagship products - Confluence and Jira.  Someone at Atlassian has a much more definitive answer, so why not share it more widely?

I suspect there is no one right answer for everybody, but there is no reason to waste disk space with files that are unnecessary and are ~90% of the backup content.

I'll report my findings, so others can have part of the answer to my question.

Kind regards,
Rich

[AnnWorley]

Hi Rich,

For Confluence and Jira, all the data is in the database and the home directory. The only reason to back up the installation directory is to preserve configuration files you may have made changes to. These vary from SSL configurations in server.xml to SSO authenticators in seraph-config.xml to java options in setenv.sh. 

To restore Confluence on a new server, for example, all that is required is to unpack an archive and point <confluence_install>/confluence/WEB-INF/classes/confluence-init.properties to the restored home directory. The home directory has the file that governs connecting to the database, confluence.cfg.xml.

These two documents are for Confluence 6.4 and Jira 7.5:

Production Backup Strategy Confluence 6.4

Backing up data Jira 7.5

I welcome any follow up questions.

Thanks,

Ann

[Rich Remington]

Thanks, Ann.

I am not current with these versions yet, but I will read up on these and send any suggestions your way.  Would I send any suggestions through this thread or in some other way?

Cheers,
Rich

[AnnWorley]

Let's continue our dialogue here on this thread, in case we discover something that will help other members of the Community.