/
Karate Test Stabilization
Karate Test Stabilization
1 root cause: Kafka Memory Limit
During running tests, Kafka pod restarts several times before and during tests. Pod restarts with OOMKilled status.
After resizing limit memory to 4096M and Xmx parameters to 2662m several modules show more stable result:
- mod-orders
- mod-circulation
- mod-ebsconet
- edge-inn-reach
Not stable modules
After resizing memory for Kafka several modules have float issues:
- data-import
- mod-kb-ebsco-java
- mod-quick-marc
More popular issues
2 root cause: Nodes large load
Before running the test adds new nodes to the cluster. First time on some "old" nodes have large load:
For resolving this issue we add a timeout before starting the test in 1 hour. After this have the next situation:
Next steps:
Like temporary solutions destroying the environment is disabled. So QA team could check the environment after running tests and investigate float issues.
More information about runs:
1 sheet - test running
2 sheet - after Kafka memory resizing
, multiple selections available,
Related content
Dojo Eureka Karate Tests Hand-off Plan
Dojo Eureka Karate Tests Hand-off Plan
More like this
Sidecar fix performance testing
Sidecar fix performance testing
More like this
PTF - Performance testing of CPU=0 for Services (QCP1)
PTF - Performance testing of CPU=0 for Services (QCP1)
More like this
[ECS] Check-in-check-out Test Report (Quesnelia)
[ECS] Check-in-check-out Test Report (Quesnelia)
More like this
Optimizing MOBIUS Environment for Stable Performance: A Comprehensive Analysis
Optimizing MOBIUS Environment for Stable Performance: A Comprehensive Analysis
More like this
Kafka KRaft Mode - Data Import with Check-ins Check-outs (Quesnelia)[non-ECS] MSK instance type comparison
Kafka KRaft Mode - Data Import with Check-ins Check-outs (Quesnelia)[non-ECS] MSK instance type comparison
More like this