Guys… check this out:
Maximum concurrent active checks …|5|
Maximum concurrent Real-Time checks…|1|
Maximum concurrent Livestatus connections…|20|
Maximum concurrent Checkmk fetchers…|13|
Maximum concurrent Checkmk checkers…|4|
I added 500 servers, just for PING monitoring. Everything else, disabled: agent, snmp. Just ping.
My cmk server died.
Fetcher usage went up to 100%.
I have this rules enabled: “Check hosts with PING (ICMP Echo Request”
Do i have to increase “concurrent active checks” ?
Will that use more memory or cpu?
Is there any guideline out there in order to set this parameters perfectly?
"If you need to ping 1000 hosts, increase concurrent active check to… X value?
“If you set 20 fetchers, set 10 active checks concurrently”
“If you need to monitor 500 hosts with SNMP, increase fetchers inmediately”
ETC
“Maximum concurrent active checks” is the correct value to change.
the number you need depends on
a) the number of checks you run actively
b) the interval
c) their runtime
d) what you want to happen if the runtime increases.
i.e.
a) 500 checks
b) every minute (guessing)
c) ping checks usually run below 1 sec, but if you have timeouts, I think they go up to 10s
d) host state is important, you might hence want to plan your active checks to cover a situation where 50% of you hosts take 10s to reply/timeout
So let’s calculate, with d) in mind, assuming 250 checks are fine, 250 might timeout:
every minute, you need active checkers for the combined run time of:
250 * 1s + 250 * 10s = 2750s
Each checker can cover 60s, so you’d need 2750/60 = ~ 46 active checkers.
Regarding memory:
the active check helpers themselves are very small (I think less than 10mb), the active check they start can be different. In your case check_ping/check_icmp are both very small, so the memory impact is probably minimal.
The calculation and memory would be very different if your active checks are i.e. check_sap_health or other checks that routinely use couple of hundred MB of memory and run for close to a minute.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed. Contact an admin if you think this should be re-opened.
