Here's the problem I'm having:
I set up Zabbix (both 3.2.6 as production, 3.4.2 as a test bed, both on CentOS 7) to get info from our switches. Right now we have 500 switch stacks with an average of 100 ports per switch stack. If I turn on all the monitoring on all the switches, it eventually shuts down Zabbix. On the test bed, I get a Zabbix queue of just over 60K, with just over 700 values processing per second. The queue from the Admin screen shows over 25K items waiting more than a minute, and 5K waiting more than 5 minutes.
I have tried tweaking settings, with mixed results, ranging from Zabbix not starting (bad) to a significant decrease in data gathering % busy (good).
Does anyone have a pre-existing my.cnf or zabbix_server.conf they can share here that would help guide me to getting this set up properly? I'd love to be able to show the bosses a "single pane of glass solution" for monitoring and not have to spend a ton of money to do so.
I set up Zabbix (both 3.2.6 as production, 3.4.2 as a test bed, both on CentOS 7) to get info from our switches. Right now we have 500 switch stacks with an average of 100 ports per switch stack. If I turn on all the monitoring on all the switches, it eventually shuts down Zabbix. On the test bed, I get a Zabbix queue of just over 60K, with just over 700 values processing per second. The queue from the Admin screen shows over 25K items waiting more than a minute, and 5K waiting more than 5 minutes.
I have tried tweaking settings, with mixed results, ranging from Zabbix not starting (bad) to a significant decrease in data gathering % busy (good).
Does anyone have a pre-existing my.cnf or zabbix_server.conf they can share here that would help guide me to getting this set up properly? I'd love to be able to show the bosses a "single pane of glass solution" for monitoring and not have to spend a ton of money to do so.