Throttle_check_thresholds: High CPU load detected

SUSE Linux Enterprise Extensions SLES High Availability Extension
1

Hello,
When running pacamaker, the CPU load is falsely detected as “High” and goes from 10 to all the way up to ~1236 and this keeps on increasing with time.
This is as indicated by the logs:

Sep 10 00:11:54 [3355] ESDNAS1 crmd: notice: throttle_check_thresholds: High CPU load detected: 1236.030029
Sep 10 00:12:24 [3355] ESDNAS1 crmd: notice: throttle_check_thresholds: High CPU load detected: 1236.020020
Sep 10 00:12:54 [3355] ESDNAS1 crmd: notice: throttle_check_thresholds: High CPU load detected: 1236.010010
Sep 10 00:13:24 [3355] ESDNAS1 crmd: notice: throttle_check_thresholds: High CPU load detected: 1236.089966
Sep 10 00:13:54 [3355] ESDNAS1 crmd: notice: throttle_check_thresholds: High CPU load detected: 1236.050049

But when check using linux top command, we get result as 98.2 % idle CPU. :

top - 13:31:01 up 1 day, 10:53, 2 users, load average: 2712.07, 2712.04, 2711.92
Tasks: 11055 total, 1 running, 11054 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.5 us, 1.4 sy, 0.0 ni, 98.2 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st

Query1: Do both these info align?
Query2: The crmd logs are flooding the pacemaker.log. Any suggestions on how to deal with it?

Any help is much appreciated

2

Its correctly detecting the systems load, it’s just not “CPU” specific as the message may lead you to believe. It’s the referring to the systems load average, which in your snippet is 2700+. So it is correctly complaining.

Rather than mask the issue I would recommend investigating what’s causing the high load average (load average is not based solely on CPU utilization). With all of the processes sleeping and the load that high I would suspect something is stuck waiting on a lock. How many of those 11,054 (wow) processes are in a uninterruptible state?

3

Thanks much!