root partition full with postgresSQL db

[flachance]

version is R81.10. The root partition on the standby Management server is full.

It seems to be caused by the postgresSQL database. On the active management server /opt/CPshrd-R81.10/database/postgresql is 3.6G, on the standby it's 41G.

Any ideas how we can clean this up and what could be the cause?

[PhoneBoy]

That would imply some sort of issue with syncing…recommend engaging with TAC to assist.

[flachance]

We already have a call with TAC. Hasn't been too helpful so far, that's why I was trying my luck here :-). We'll keep working with TAC. I'll post any new details I get here.

[the_rock]

Did they give you any advice so far? I see what phoneboy means by sync issue, that definitely makes sense...I also see what @Bob_Zimmerman is suggesting, sounds like the fastest way to solve this problem.

[flachance]

Until just about now, the advices were to clean up unnecessary files and expand disk space. We informed them today on what we found regarding postgressql seemingly being the issue and we just received instructions to install Take 75 (we're at Take 66). They referenced sk178889. So we'll look at that, although I suspect installing a JHF when the root partition is full might be problematic 🙂

[the_rock]

Honestly, if I were you, I would NOT touch that at all. Anything to do with database (regardless if its gw or mgmt, SPECIALLY mgmt server), deleting those files can cause corruption beoynd repair. As @PhoneBoy , you are better off getting TAC involved and confirm with them. One thing I would do beforehand is maybe compare that content of that dir on both servers and see whats actually different. I get that space is way less on standby, but real question is why.

[Bob_Zimmerman]

Not sure on the cause. You would need to call the TAC.

As for cleaning it up, I would just flatten the standby management and rebuild it. Inconvenient, but pretty quick, and guaranteed to work with no residual gunk left over from before rebuilding. Secondary managements should be as transient as firewalls.

[flachance]

Yes it looks like rebuilding it with the latest JHF might be the fastest/easiest