Hello Dear ,
We have a zabbix stack with the following architecture Three front end load balancers Two Zabix servers as HA A physical PostgreSQL database 10 proxy servers in different zones (3 proxies for Windows server zones - 3 proxies for Linux server zones - 2 proxies for network equipment - 2 proxies for virtualization infrastructure).
In this collection, a total of 6100 hosts and 742,000 items are monitored, and 94,000 inactive items are disabled and 143,000 items are unsupported.
Description of the problem Queue value in the proxy has suddenly reached 60 thousand items and all Windows agents stop working. To solve this problem, by killing the Zabbix server, the second server will be activated and after about 5 minutes, the queue will be empty and the problem will be solved.
With the checks we do. Changing the poller parameters in the server and proxy, as well as the timeout, this problem still remains, and generally this problem happens once a day at different hours.
Checking the database does not show any problems. Before the incident, all items were working without problems.
At the moment of the incident, the server log shows that the processing is done on the server items and does not show any problems (it seems that the processing speed is greatly reduced). In terms of resources, there is no limit in the service and there is still a problem with adding resources It seems that in a certain period of time, one/or several items suffered an error in pre-processing, and this leads to the occurrence of a disturbance that has not been successful in finding this item so far. I would be grateful if friends have a solution in this regard to let me know.
Our zabbix version is : 6.0.12
We have a zabbix stack with the following architecture Three front end load balancers Two Zabix servers as HA A physical PostgreSQL database 10 proxy servers in different zones (3 proxies for Windows server zones - 3 proxies for Linux server zones - 2 proxies for network equipment - 2 proxies for virtualization infrastructure).
In this collection, a total of 6100 hosts and 742,000 items are monitored, and 94,000 inactive items are disabled and 143,000 items are unsupported.
Description of the problem Queue value in the proxy has suddenly reached 60 thousand items and all Windows agents stop working. To solve this problem, by killing the Zabbix server, the second server will be activated and after about 5 minutes, the queue will be empty and the problem will be solved.
With the checks we do. Changing the poller parameters in the server and proxy, as well as the timeout, this problem still remains, and generally this problem happens once a day at different hours.
Checking the database does not show any problems. Before the incident, all items were working without problems.
At the moment of the incident, the server log shows that the processing is done on the server items and does not show any problems (it seems that the processing speed is greatly reduced). In terms of resources, there is no limit in the service and there is still a problem with adding resources It seems that in a certain period of time, one/or several items suffered an error in pre-processing, and this leads to the occurrence of a disturbance that has not been successful in finding this item so far. I would be grateful if friends have a solution in this regard to let me know.
Our zabbix version is : 6.0.12