Distributed Monitoring Backups

[mmccaugh (Michael McCaughan)]

I have been searching forums but the info I have found is not making sense to me so far.

I have a distributed monitoring setup with a SPOG server, and then 6 distributed pollers, I have WATO disabled on the remote sites so all configuration is done via the SPOG.

Where I am stuck is how i set up backups on my remote pollers? What I read in one thread is the backup from the SPOG should contain all devices, but doing a tar tvfz against it I only see the site on the SPOG and none of the remote sites/devices in the archive.

Now looking at the archive it seems like it is just the content of /opt/omd/sites/ so I was not sure if I could simply wrote a quick bash script to tar this directory on each poller and pop it into cron?

Obviously the issue here is this might ‘look’ easy, but until I’m in a position of needing to restore a backup I have no idea if I am missing something! So just been reading and trying to find the ‘proper’ process to do a full backup of a distributed topology so I am covered in the event I ever lose a single poller, or the SPOG server…

[paulosantanabr (Paulo Santana)]

For Full Backups I would recomend using omd backup in each server and storing the file in a safe location. It will contain everything to allow the recovery in a disaster cenario.

[mmccaugh (Michael McCaughan)]

Ahhhh OK I see what you mean now!

omd backup [SITE] [-|ARCHIVE_PATH] Create a backup tarball of a site, writing it to a file or stdout

WATO is disabled but I can still do it via the cli with the omd binary!

Testing it now and it appears to be working, so there was a super obvious solution I was overlooking

Cheers!

[Anders]

We run rsync to another backup site (for each Checkmk Site/Poller) every hour so we don’t loose data and can perform a failover. We also use the exact same method as @paulosantanabr suggested before we do upgrades so this works fine.

Bear in mind that you might have other things outside the side that you want to backup (Apache configs etc) - They won’t be included in the tarball.

[mmccaugh (Michael McCaughan)]

Yep I have all that set already! I have a standard build image/process as well as a system backup so my only real concern was the application specific data (Which is changing daily lol)

Solution was perfect and I have already tested, and I do have a backup target for each site (Which I rsync backups to already) that I have added these to!

Thanks again!! Ironically the CLI backup is a BETTER solution than via WATO as now instead of a process to “check” if the backup exists, I just add the transfer to the end of the backup script lol… So I came out of this with a better plan

[Anders]

yea quoting ppl here seems to be just… well the intelligent ones