Zabbix proxy v7.0.14 preprocessing overload after upgrade

[vad-valrc]

Hello,

We are running our Server in version in v7.0.8 and we've update our proxies in v7.0.14. After this update, many of our proxies trigger an alert : Zabbix proxy: Utilization of preprocessing manager processes over 75%

We tried to increase number of preprocessor without any result. Then we believed we found the solution by updating HistoryCacheSize to a higher value. It solve the issue but for a limit amount of time.

In logs we have many error of this kind but not on all affected proxies so maybe not related:
1099958:20250619:110037.110 received value identifier 13332074 from proxy 20576 is lower than the last processed value identifier 13332075

This preprocessing overload generate a lot of cpu usage too. I think we'll have to rollback this version. I tested v7.0.13 yesterday without any issue and update this morning but didn't check there was a new version.

Is it related to the fact server is not aligned? is there new parameters to setup to avoid this load ?

Thank you

[ysus]

Fixed in 7.0.15

ZBX-26588

https://www.zabbix.com/rn/rn7.0.15

[vad-valrc]

Thank you for sharing

[beemo]

Hi, same problem here i upgraded Monday the server in v7.0.13 all was ok, as i saw an update of agents in 7.0.14 then i upgraded the server today in 7.0.14 after upgrade en reboot => alert "Utilization of preprocessing manager processes over 75%"

[beemo]

Here some graphs/values showing the problems
Zabbix server: Utilization of preprocessing manager internal processes, in %: 99.109 %
Preprocessing queue 0 (before) now growing to 368
Discovery queue 0 (before) now growing to 750
server: Queue => constant at 105 (before some spikes)
Calculation process at 99%

I manage to recover the situation trying some changes in /etc/zabbix/zabbix_server.conf

### Option: StartPreprocessors
# Number of pre-forked instances of preprocessing workers.
# The preprocessing manager process is automatically started when preprocessor worker is started.
#
# Mandatory: no
# Range: 1-1000
# Default:
# StartPreprocessors=3
StartPreprocessors=5

systemctl restart zabbix-server.service

Graphs and values returned to "normal" as before.

[vad-valrc]

Hello,
In our side we rollback zabbix proxies to 7.0.13 and everything went back to normal but now I'm upset to upgrade to this version. The number of preprocessor on proxy side didn't solve the solution and we had a huge CPU load on all migrated proxies.
I'm wondering if this new agent feature require some specific setup on server and proxy ?

[beemo]

This morning check indicate a problem, the queue problem seams to be related to a windows Hyperv system
" Zabbix server: More than 100 items have been missing data for over 10 minutes" and i have graphs indicate queue problem too

[ysus]

Fixed in 7.0.15

ZBX-26588

https://www.zabbix.com/rn/rn7.0.15

[vad-valrc]

Thank you for sharing

[beemo]

Hi, thanks to, i update this week-end server side, and some agents like hyperv (update in .msi version 7.0.12) the saturation is gone after the agent upgrade

[muelli]

I have 1 platforms where the problem appears in 7.0.15 but 7.0.14 was OK and 1 platform where 7.0.14 showed this error but 7.0.15 runs (for now).....
Now the zabbix-server part died without notice. Downgrade does not work, it simply dies again after a couple of minutes. the update seriously messed something up