Proxy mnt/database space grow too fast

[Jeff54]

Hi,

I face a problem on one of my proxy.

I have a Zabbix server and 3 active proxies dispatch on 3 differents sites.
On one of them, I can see the location of the data storage mnt/database who grow too fast. Last time we had to backup the VM to make it work again.



I have the same configuration on all my proxies and I don't understand what can be the reason of the problem for this one.

Did someone encounter the same ? Or have an idea to what I can do/check ?

I'm new to Zabbix and I'm completely lost with this problem.
Thank a lot.

Regards

Jeff

[Jeff54]

Hi,

After I use a snapchot to restore the proxy, the storage start to shrink database like it's configured I think.
I don't explain why it's good now...


I will close this case and I hope it won't be a problem again in the futur.

Jeff

[cyber]

What do you use as DB on proxies?
Easiest would be to go with sqlite3... Never had issues with it. Specially easy to clean up, just by deleting the DB file, which willbe recreated after proxy startup..
But I guess you have something else there... MySQL/Maria?
Proxy DB should not contain much more data, than what it has gathered and not yet sent away (plus data about monitored hosts themselves). So I really doubt you should have tens of gigs of data... Probably should dig into DB logs there. some issues with cleaning up maybe...

[Jeff54]

Hi,

Thanks for you answer.
We use PostgreSQL, Did you know if it's better to delete files on the mnt/database directory or modify a parameter on the zabbix_proxy.conf.
Because I don't know that I can modify to improve the config.

Regards

[Jeff54]

Hi,

Nobody face an issue like that ? I have to restore my Proxy every two week and that's very problematic.
It's in production and I think to recreate one but that will be complicated in my company.

Regards

[cyber]

I have not use PG in proxies, so this is a unfamiliar territory..
Seems like housekeeping is not working at all... check logs, its there anything about housekeeper.. ?

[Jeff54]

Hi everyone,

I clone the VM of one of a good proxy and use it to replace the one where I face this problem but I have the same issue.
So the parameters like housekeeping for example seems to be good

The mnt/database has a directory 16385 (don't understand why this name) with a lot of files, they are all growing until they reach the size of 1Gb. Then an other is created, etc.
They all have a name like 17492, 17492.1, ... and after a 17510.

The inside is non readable and I am completely lost with this problem.
I made a script to erase some but sometimes it makes crash th Postgres DB.

If someone has an idea, I would be very grateful.

Jeff

[Jeff54]

A brief return of what happened during the night.
The connection completly crashed with the database. When I check the evolution of the storage on the /mnt/database, it raised very quickly but I don't understand why...

Here is an other capture of the logs.
We see that the shutdown start at 02:42, may it be a ressources problem ? I don't think it's normal to store 15Go or more in less than 6 hours on a proxy...


Regards,

Jeff

[Jeff54]

Hi,

After I use a snapchot to restore the proxy, the storage start to shrink database like it's configured I think.
I don't explain why it's good now...


I will close this case and I hope it won't be a problem again in the futur.

Jeff