queue problem

[perun.84]

It's already done...

[mortuletti]

Good!
Hope, queue will not grow and system works as expected.
Regards,
Alexander

[perun.84]

But queue is growing. We'll see. I hope that it'll be OK. Thanks

[mortuletti]

If it is growing, need to continue troubleshooting..
So, most of delayed Items is "Description" and "Alias of interface" (text information)

In latest data do you have some data collected for problematic Items?
Is it related to all devices or just some?
Is it possible to reduce number of suspects? (network segment or device type)

Br, Alexander

[perun.84]

It's going with some of devices. But the other items of those hosts are collecting regularly (Outgoing and incoming traffic are collecting every minute..)

[mortuletti]

Try to list all devices with problems and check type and model. Is it different or the same?
When last successful check was done for each problematic Item? Could be, some change was done just after that.
Then, could be good to check MIB responsinility. Check/install net-snmp net-snmp-utils on Proxy and try to get information directly using snmpwalk and snmpget.
Does devices respond to snmpget requests? (this is actually the same way as zabbix get the data)

If problematic devices can send SNMP traps, try to reconfigure Items to receive SNMP traps. This can be as possible solution for this type of items.
SNMT Trap config
Br, Alexander

[perun.84]

Models are various. I've tried to get missing information using snmpwalk, and device is answering well. I couldn't do it by traps because of network devices CPU and memory..

[mortuletti]

Hi!
If using snmpget command you are getting value without huge delay, check what is wrong with Item key. Zabbix getting information the same way.
What is a time/date of last successful check? Was something changed right after?
Br, Alex

[perun.84]

Last check was over two days ago. Zabbix set to check those items once a day. Snmpwalk works without significant delays...

[mortuletti]

Please, try to switch off "Use bulk requests" on SNMP interfaces for problematic hosts.
Could be, some devices does not support for this option.
Br, Alex

[perun.84]

I should find way to do it at all devices once. Maybe I may to do it by sql query directly in db...

[mortuletti]

Please, first try to do it on most problematic hosts manually.
If this will helps, I'll help with script for all devices.
To make sql query, need to filter out only devices where this change is needed, not for all, I guess.
Br, Alex

[perun.84]

OK. I'll try. I've changed database (improt from friday-s backup) and after that last checks and graphs show strange times ( with 7 hours 32 minutes delay). On the other proxy, everything is ok.

[mortuletti]

Hi!
Can you share one of examples?
Br, Alexander

[perun.84]

Absolutely. Image is below. Time is 05:06 but real time was 12:38...