Sorry for the images missing in the above description. Please find the correct one below

Please find the parameter configuration which is same for both the proxies and the memory free also:

Please find the memory util of prx-03 :

please find the diaginfo of prx-03:

Could you please be so kind and identify exact process that is consuming memory ?

Hello please find the snapshot of list of processes with memory util

Hello vso have attached the list of processes with their memory utils. Can you please check and help me out?

Please let me know if any additional information is required.

Please try setting environment variable for Zabbix proxy service and restart

export MALLOC_TRIM_THRESHOLD_=134217728

Hi vso, have tried the command you shared. And checked for 48hrs there is no change. The memory util went high to 95% gradually.

Have restarted the VM as well and entered the command you shared. But still same behaviour.

Hi vso, tried using the libjemalloc module in "LoadModule" parameter in proxy config file. When restarted have got this error : 'cannot load module "/usr/lib64/libjemalloc.so.2" : /usr/lib64/libjemalloc.so.2 : cannot allocate memory in static TLS block' in zabbix_proxy.log

It should have been loaded with LD_PRELOAD such as:
LD_PRELOAD="/usr/lib64/libjemalloc.so.2" ./sbin/zabbix_proxy -c /etc/zabbix/zabbix_proxy.conf --foreground

Or just:
export MALLOC_TRIM_THRESHOLD_=134217728
./sbin/zabbix_proxy -c /etc/zabbix/zabbix_proxy.conf --foreground

Current suspicion is that there are event log keys that send too much data, is VMware used ?

Yes we are monitoring VmWare, but vmware logs monitoring is disabled.

There seem to be similar issue but with higher consumption:
ZBX-26301

Could you please be so kind and double check with following query ?

select key_,status from items where key_ like 'vmware.eventlog%' and status=0;

Hello vso, have checked on the vmware eventlog items. Those are not present as I have removed them from the template itself.

Also, I suspected network discovery. I have 18 network discovery rules with /22 subnet and thier respective discovery actions.

I have disabled all these network discovery rules for 24hrs, but still the memory utilization is reaching 98% and crashing.

Have tried LD_PRELOAD command that you shared, which is making the proxy in activating stage but never coming to active state.

Then i have restarted the VM then the proxy started and acting same reaching 98% memory util in 24hrs.

There seems to be no issue with preprocessing manager, please lower poller count (it looks like there are hundreds in latest screenshot), since most of the checks can be done by async pollers you probably just need few StartPollers and increase only if you start to see queue increasing or poller utilized too much.

Ok sure, will reduce to the necessary and will monitor for 24hrs.

Hi vso, have reduced the number of all pollers and could see there is decrease in the memory utilization. Now it seems to be constant util at 70% since 48 hrs.

Now I have further decreased the pollers and will monitor further.

Thank you, closing for now