Zabbix 7.0.3 - 100% memory utilization after some time

[Study9059]

Hello,

Lately, memory utilization of Zabbix went completely out of control needing a restart every 3 weeks on average. I'm using 7.0.3 and the problem is here since the upgrade to 7.0 i guess.

Here are some pics.



You can see that this is zabbix_server eating all the free memory day after day. Memory usage from top command :



Not big deal but it is kind of annoying to deal with this bug, if that's a bug.

Thank you.

[kamil1]

Hi,
Can you clarify couple of things:

• Are there any configuration changes between the upgrade to 7.0.3?
• Have you checked whether you are correctly handling triggers/events and there are not increasing over time?
• Are you using any external scripts or custom templates that may be consuming memory?
• Have you checked the Zabbix server logs for any errors or warnings related to memory usage?
• What is the server’s hardware configuration (RAM, CPU)?

Thanks

[Study9059]

Hi,

Thank you for your time, i checked all your points :

• No configuration change whatsoever between 7.0 and all minors after that
• Correctly handling triggers and events : i guess (i mean i never changed my way of doing things and yet memory started increasing its consumption)
• Not using scripts from the server itself, but i made some scripts running on a couple of hosts that are managed by the zabbix agent (very light scripts)
• Logs are not relating any errors
• Server is a RHEL9 VM with 4 CPU and 24 GB of RAM (at first it was 8GB but the memory consumption made me increase it like hell)

What is very concerning is that a reboot fixes the issue temporarily, it's the reason why i think it has nothing to do with the number of metrics/triggers/events.

[cyber]

Here comes the "support person question".. Have you thought of updating to 7.0.4.. ?

[Study9059]

Hi,

Yes, we're currently using the 7.0.6, but unfortunately the problem is still present.

Thanks

[cyber]

Which server process is that in top? (press "c" to see longer command line in top)
All I can see, that you have all-in-one instance... Maybe it just needs some more resources? Load over 5 there (how many CPUs?) .. just 24G (ish) memory for all things (server, DB, UI) ... maybe you should split them up to 3 separate hosts and give a bit more resources to DB...

[Study9059]

Hi,

It seems that, according to top, this is zabbix_server with this command line :

3881 zabbix 20 0 10,2g 6,9g 35584 S 0,7 29,6 65:43.79 /usr/sbin/zabbix_server: preprocessing manager #1 [queued 319, processed 324 values, idle 5.001512 sec during 5.009956 sec]​

Yes, using an AIO with Postgres on the same host. Currently it have 4vCpu and 24 gig of ram, i'll push it to 16CPU and 128Gig to see how it goes.

[cyber]

it is preprocessing manager, which seems to have lot of things to do .. Maybe adding some preprocessing workers makes manager also happier... (starting from 7.0.6 default is 16, before that, just 3...)

[Study9059]

Hi,

After some days, it seems that boosting the VM to 16CPU and 128Gig of memory resolve the issue. Since my last message, the server is idling at 2.2% as you can see below :



That said, 128gig is way over the real needs i would like to decrease it.