-
Bug
-
Resolution: Unresolved
-
Undefined
-
None
-
False
-
-
False
-
None
-
rhel-se-kernel
-
-
Beaker jobs on rhel-9.6:
https://beaker.engineering.redhat.com/jobs/12267834
https://beaker.engineering.redhat.com/jobs/12267859
https://beaker.engineering.redhat.com/jobs/12268980
Even there is no any failure it has happened.
Main Problem Analysis
1. Perf Interrupt Processing Delay
The logs show numerous messages like:
perf: interrupt took too long (2508 > 2500), lowering kernel.perf_event_max_sample_rate to 79000
This indicates that perf subsystem interrupt processing time exceeded the expected threshold (2500 microseconds), forcing the kernel to continuously lower the sampling rate.
2. Time Gap Analysis
From the logs, I can see:
- Test #95: 07:28:33-07:28:35 (Duration: 68s)
- Test #96: 07:34:38-07:34:40 (Duration: 359s)
- Test #97: 07:37:24-07:37:26 (Duration: 160s)
Time intervals:
- #75 to #76: ~16 minutes gap
- #95 to #96: ~6 minutes gap
- #96 to #97: ~3 minutes gap
Potential Root Causes
1. High System Load
- CPU resources heavily consumed by perf event processing
- Interrupt processing delays causing overall system performance degradation
2. PMU Resource Contention
- Multiple perf tools simultaneously accessing Performance Monitoring Units (PMU)
- Insufficient hardware counter resources
3. Memory Pressure
- Memory dump at log end indicates potential memory pressure
- Large perf data buffers may consume significant memory
4. Test Framework Issues
- restraintd daemon may have additional cleanup/preparation work between tests
- File uploads (Uploading resultoutputfile.log) may cause delays