Using Zabbix 6.2.9 on Rocky 8. I tried setting StartPollers=56 and StartPollersUnreachable=10, but the message still does not go away. What should we change?
Ad Widget
Collapse
Can't Get Rid of This: Zabbix server: Utilization of ODBC poller processes over 75%
Collapse
X
-
Tags: None
-
Increase the number of ODBCPollers
/etc/zabbix/zabbix_server.conf
## Option: StartODBCPollers
# Number of pre-forked ODBC poller instances.
#
# Mandatory: no
#Range: 0-1000
# Default:
StartODBCPollers=1
Others parameters: https://www.zabbix.com/documentation.../zabbix_server
Wellington👍 1 -
I tried setting it to 6 but it didn't make the message go away. If I set it to anything higher, CPU utilization goes to 90%. It's an 8-core VM. There's no reason why ODBC polling processes should require so much power.Increase the number of ODBCPollers
/etc/zabbix/zabbix_server.conf
## Option: StartODBCPollers
# Number of pre-forked ODBC poller instances.
#
# Mandatory: no
#Range: 0-1000
# Default:
StartODBCPollers=1
Others parameters: https://www.zabbix.com/documentation.../zabbix_server
WellingtonComment
-
If you monitor ODBC you will need to increase.
Confirm how much Latest data is being used on your Zabbix Server.
Remember: After increasing, always restart the Zabbix-server service and wait some time for the entire queue to be processed.
Confirm which process is using the most CPU.
And I didn't forget to search and read:
Bugs: https://support.zabbix.com/browse/ZB...~%20%22odbc%22
Evaluate the performance of the database and other Zabbix pollers.
And version 6.2.x is no longer supported, consider upgrading to 6.4
Wellington
Comment
-
Comment
-
We've been doing ODBC polling of about 400 MySQL (Percona) databases for a year and it has been working. I understand that polling ODBC requires more power than not polling it, but I would think that doubling it from 4 cores to 8 and restarting the service would be enough to get rid of that message.If you monitor ODBC you will need to increase.
Confirm how much Latest data is being used on your Zabbix Server.
Remember: After increasing, always restart the Zabbix-server service and wait some time for the entire queue to be processed.
Confirm which process is using the most CPU.
And I didn't forget to search and read:
Bugs: https://support.zabbix.com/browse/ZB...~%20%22odbc%22
Evaluate the performance of the database and other Zabbix pollers.
And version 6.2.x is no longer supported, consider upgrading to 6.4
Wellington
We are indeed using the mariadb connector for ODBC.
Hopefully 7 will do better.Comment
-
Hello,
How much ODBC pollers should have for monitoring around 20 DB with ODBC Poller because I am reaching this trigger: Utilization of ODBC poller processes over 75%, when 2 of them goes down. In that case the trigger is raised and I am receiving from all DB's the trigger "Failed to fetch info data for last 5 minutes"
So how much is suggested for 20 DB monitored by ODBC?
#StartODBCPollersComment
it will push 6.4 to unsupported anyway. At least it will be LTS and comes with all the stuff 6.4 already has... and has longer support.
Comment