Proxy host items queued with next check in the past

[elvar]

Hello,

I have 11 switches at one remote site being monitored by a Zabbix Proxy and some of these switches keep having issues where after first getting linked to the template they pull some data, but then very soon stop pulling data. The debug logs of proxy and server don't show any actual errors, but I end up with a few hundred queued checks all with "Next Check" time/date stamps in the past.

If anyone could help me out on this one I would greatly appreciate it. Non-proxy host items are not having the same type of queue issues if that helps.

Zabbix Server & Proxy 1.8


Regards

[exkg]

But how many time in the past ? Is it in '1969' ?

All new items have 'next check' before unix timestamp ... if your items ALWAYS have this date/time means that this items NEVER had data.



[]s,
Luciano

[elvar]

Doh

But how many time in the past ? Is it in '1969' ?

All new items have 'next check' before unix timestamp ... if your items ALWAYS have this date/time means that this items NEVER had data.



[]s,
Luciano

Well, I think I may have finally figured out part of it, but there still remains some confusion. I can see that lots of these queued items that have been queued for 2 hours or more are ports that have no active link at the moment. But what I do not understand is that I have a Simple Check (icmpping) for those same switches and that is queued at 2 hours + as well. Why wouldn't the (icmpping) check keep happening considering it's set for a 30 sec interval? Does a lot of queued items disable a host?

[exkg]

Hi,


How many items per second have you ? If this number is 'high' maybe you need think increase your number of pollers.

And about your log file ? Some information about this scenario ?


[]s,
Luciano

[elvar]

Hi,


How many items per second have you ? If this number is 'high' maybe you need think increase your number of pollers.

And about your log file ? Some information about this scenario ?


[]s,
Luciano

Currently my server's required values per second is 55. I'm not seeing any indication that the server or proxy is unable to keep up with the volume. Also like I mentioned, I have no queue issues with Zabbix Agents or simple checks which initiate directly from the Zabbix Server, only with hosts monitored by the one proxy I have in use. It also isn't all switches being monitored by the proxy either, only the switches that have a lot of ports without active links because many of the computers are turned off at this time. Those switches with many ports offline start queuing up and then I see my simple check get stuck as well for some reason. I have debugging enabled on both the zabbix server and the zabbix proxy. All I keep seeing in the zabbix proxy log related to one of the hosts in question is the following...

'tail -f /var/log/zabbix/zabbix_proxy.log | grep MyHost'
"host":"MyHost-Switch-Room21-2900-48G",
"host":"MyHost-Switch-Room21-2900-48G",
"host":"MyHost-Switch-Room21-2900-48G",

When I tail the log I keep seeing that but that's about it. I'd very much like to know why it doesn't keep trying to do the "icmpping" check. I see the icmpping check stuck in the queue but shouldn't it keep trying?

[elvar]

Solved

Solved... I think

Just like I did with my Zabbix Server mysql database, I applied the recommended tweaks in http://www.zabbix.com/forum/showthre...sql+consultant to my Zabbix Proxy mysql database and haven't noticed the queue problems. Upon further inspection I had found errors about tables being full that I had initially missed earlier today when troublshooting. I really hope this helps someone else out.


Cheers!