Required performance / Values processed

[Lennyroquai]

Hi,

Recently i started having some strange thing with my Zabbix :
I have a lot of "xxx failed: first network error, wait for 15 seconds"
and a few seconds later : connection restored
And, a small hole in graph.

For the moment, only Agents and JMX are concerned (SNMP is working fine)

I have Zabbix 2.4.1, one server acting as Zabbix server, and another one as database.

When i check Zabbix Health, it says :
- Required performance of Zabbix server, new values per second : 389.28
- Values processed by Zabbix server per second : 326.47

All internal processes are under 5%
Gathering processes : under 25%
CPUs < 15% (on DB and Server)

I am monitoring : 180 virtual machines, 16 physical servers, 16 switchs, 8 routers, 32 Java servers, 8 PDUs....
And my Zabbix_server.conf is heavily tuned (Cache, pollers, etc...)
Of course, connectivity has been tested.

And here's what logs look like (Debug level 3)


Does someone has a clue about why this is happening ?

Many thanks

[tchjts1]

You can try to increment your Timeout= value on your Zabbix server in zabbix_server.conf if it is still at the default of 3. Maybe try 15 or 20 and then restart your Zabbix server process.

I know you say your internal processes are good, but I would be interested in seeing the 3 default graphs which show all those procs, with a 1 day (24 hour) view, as per the bottom of this post: https://www.zabbix.com/forum/showthread.php?t=41219

[Lennyroquai]

Hi tchjts1,

Here's my zabbix.conf :

StartPollers=80
StartIPMIPollers=5
StartPollersUnreachable=3
StartTrappers=16
StartPingers=12
StartHTTPPollers=4
StartTimers=4
StartJavaPollers=6

StartVMwareCollectors=4
VMwareFrequency=30

StartDBSyncers=4

Timeout=30

An here's the graph requested (sorry, i didn't see the post talking about it)







Other infos :
My queue seems to be clean : the only strange thing is that, sometimes 81 items seems to get stuck in the queue.
If i restart the server, they dissapear... but then come back a few minutes later... and stays in the "More than 10 minutes" column.

My issue seems to target only Windows machine (2.4.0 agent) and if i checked the log : it's always the same machines looping in it. (but they are on different network)
I've restarted the agents yesterday, but nothing changed.
I will install the new agent to see if it solve the issue.

Thanks

[tchjts1]

Your processes all look fine.

On the agent where you are seeing that issue, in zabbix_agentd.conf is also a Timeout= value. Try incrementing that and restart the agent.

One other thing to check on that one - be sure you installed the correct agent - 32bit or 64bit.

[Lennyroquai]

Hi,

I've tried :
- Set the Timeout Value to 30s
- Upgrade the agent to 2.4.1
- Downgrade the agent to 2.2.1

Nothing new...
Currently, the agents with have the issue are all 32 Bits, (Windows 2003 Virtual machines)

[Lennyroquai]

Damn... I finally found what was wrong...

The interval of an item, in my "Windows Template" was set to 1.

I don't know why and when this happened, but once a set it back to 60 : The issue stopped !



Anyway, thanks for the help.