-
Story
-
Resolution: Done
-
Highest
-
None
-
None
Recently we had the feeling that master gate was broken because several smoke tests were reporting FAIL.
However reverting on different OOM Master gates show that it was not really the case.
The different pythonsdk based tests (basic_onboard, basic_vm, basic_network, basic_cnf) are failing for different reasons, sometimes we may observe one or several errors on the same gate/daily, sometimes, everything is PASS.
As a consequences the smoke suites do not offer a deterministic way to evaluate the patch and the quality of the Daily.
So far the different reasons already observed can be:
| Componet | Description | JIRA | Status |
|---|---|---|---|
| VID | VID does not start properly When building object (LoB, OwningEntity => 500 from the VID |
https://jira.onap.org/browse/VID-960 | |
| SO | instantiation stuck at 36% Cassandra and MariaDB Galera exception seen in so-bpmn Note that there is no timeout, the instantiation request to the SO got a 200 but everything is stuck |
https://jira.onap.org/browse/SO-3584 | |
| SDC | resources available then not found on certification attempt It seems that serialization improves the situation but // is still working in Guilin |
https://jira.onap.org/browse/SDC-3508 | |
| SDC/Tester | SDC package seen as empty, no version can be found tester was retrying wildly (patch submitted by Michal to stop stressing the SDC) leading to a 500 on SDC |
OK | |
| SDC | SDC package seen as faulty ("ZIP_FILE_CREATION__MISSING_ENTRY_DEFINITION","message":"Failed to create CSAR file from ToscaServiceModel,missing mandatory EntryDefinitionServiceTemplate ") | ||
| Tester/SDC | SDC reports a Malformated package patch submited by Christophe to add the manifest in the test packages we are using (SDC check format evolution for Honolulu) |
OK | |
| SO /SDNC | SO /SDNC :{"requestState":"FAILED","statusMessage":"STATUS: Error Source: SDNC, Error Message: Error from SDNC: No service logic active for generic-resource-api: 'service-topology-operation' FLOW STATUS: UnassignServiceInstanceBB has failed. TASK INFORMATION: Last task executed: Call SDNC RETRY STATUS: Retries have been exhausted. ROLLBACK STATUS: Error from SDNC: No service logic active for generic-resource-api: 'service-topology-operation'","percentProgress":100,"timestamp":"Thu, 11 Mar 2021 11:31:15 GMT"}}} ) => basic_network|vm|cnf FAIL | https://jira.onap.org/browse/SDNC-1512 | |
| SO | SO: 502 on SO request (Client from http://so-bpmn-infra.onap:8081//mso/async/services/WorkflowActionBB failed to connect or respond) |
In parallel, the recent tests on the daily Guilin seems always better than the one on the Daily master.
Assuming that the tests used are the same (xtesting-smoke-usecases-pythonsdk:master and xtesting-onap:master (to launch them), Master seems to be more exposed to the problems listed above.
One of the tester main evolution in master was dealing with the fact to get customized ONAP exceptions. Before this change (onapsdk 7.4), at the tester level we were catching the exception widely, which may have hidden some of the issues we currently met
The smoke tests have always been realized in //, it was working reasonably well and save CI time. However it complexifies the troubleshooting. As a consequences an offset has been added to avoid starting at the same time but no effect has been really obsvered, the offset is maybe too small.
- mentioned in
-
Page Loading...