Can db connection errors increase unreachable poller process processing time?

[mjhs6967]

When a db connection error occurred due to a network failure, the processing time of the unreachable poller process increased highly.

Could this affect the unreachable poller?

Plus, how exactly does the unreachable poller process work?..

[max.]

Hello,

Unreachable pollers are responsible of host availability. So if there is some kind of network issue its normal that utilization will increase dependening of the number of hosts affected and the number of pollers.

How it works:

If a host becomes unreachable, the poller checks every X seconds (default 15) if there is a response from the host.
That process is repeated for X seconds (default 45) until the host becomes unavailable.
When the host is unavailable, the host is checked for aviability every X seconds (default 60).

This values can be modified in the zabbix server configuration file, but i think the default state is good enaugh.

### Option: UnreachablePeriod
# After how many seconds of unreachability treat a host as unavailable.
# UnreachablePeriod=45

### Option: UnavailableDelay
# How often host is checked for availability during the unavailability period, in seconds.
# UnavailableDelay=60

### Option: UnreachableDelay
# How often host is checked for availability during the unreachability period, in seconds.
# UnreachableDelay=15


Regards

[mjhs6967]

First of all, thank you for your answer
But during the time when there was a network issue, there were only 6 network error logs in zabbix.
(the network monitoring most hosts and the db connection network are separated.)

Does the value of unreachable poller process seem reasonable when looking at the error logs of the 6 cases?

zabbix server log
---
8571:20220418:084345.948 SNMP agent item "CPU_Util" on host "#" failed: first network error, wait for 15 seconds
8555:20220418:084424.457 SNMP agent item "CPU.Utilization" on host "#" failed: first network error, wait for 15 seconds
8597:20220418:084824.183 SNMP agent item "Memory.Utilization" on host "#" failed: first network error, wait for 15 seconds
8603:20220418:085157.421 SNMP agent item "CPU_Util" on host "#" failed: another network error, wait for 15 seconds
8498:20220418:085345.129 SNMP agent item "CPU_Util" on host "#" failed: first network error, wait for 15 seconds
8601:20220418:085415.403 SNMP agent item "CPU_Util" on host "#" failed: another network error, wait for 15 seconds
---

value of unreachable poller process
---
2022-04-18 08:53:10 4.1937
2022-04-18 08:52:10 21.5273
2022-04-18 08:51:10 54.9375
2022-04-18 08:50:10 66.1089
2022-04-18 08:49:10 66.4583
2022-04-18 08:48:10 68.1416
2022-04-18 08:47:10 65.6142
2022-04-18 08:46:10 54.9441
2022-04-18 08:45:10 33.7657
2022-04-18 08:44:10 11.3683
2022-04-18 08:43:10 3.5664
---

[cyber]

If ordinary poller fails to poll a value for item, then UnreachableDelay kicks in and its tried again in that time until UnreachablePersiod expires... so having 15 and 45 there by default means it is checked 3 times... if it's still not reachable, then host is declared Unavailable and polling is disabled. Its polling is given over to unreachable pollers, which try it in UnavailableDelay persiod. When they finally manage to reach it, its given back to normal pollers.

[mjhs6967]

Thank you for your help.

Can you check the above comment?
I don't think the value of unreachable poller process is a reason to go up that much.

[cyber]

If you have overall network issue, then unreachable poller usage can go and will go up.
How many of those pollers you have? Whats the timeout value? if you have long timeout, like 20s (default is 3 I think), then your unreachable poller will wait for that timeout, if you have not too many unreachable pollers, they will all be occupied, usage goes up...

[mjhs6967]

StartPollersUnreachable=10 and Timeout value is set 4 (default is 3) Does the value of unreachable poller process indicate the time taken while the host is declared to be unreachable?

[cyber]

https://www.zabbix.com/documentation/6.0/en/manual/appendix/items/unreachability

Here is described how it works...