Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

Duration of the same test may vary a lot and globally increase over time

Description

During the stability campaign test, we detect that the test duration of a simple test was varying a lot from one instantiation to another.

see https://docs.onap.org/projects/onap-integration/en/latest/integration-s3p.html#integration-s3p

at the beginning it is pretty stable but progressively we have duration from 3m to 30m for the same code execution.

Preview unavailable

Apparently this phenomenon started in Frankfurt (Common DB?, refactoring?)

It would be nice to investigate a little bit to try to understand and reduce the drift

assuming that the tests is OK, why does it sometimes look stuck more than 25m

Environment

Attachments

1
  • 10 Dec 2020, 07:28 AM

is blocked by

Confluence content

mentioned on

Activity

Show:

Former user June 16, 2021 at 2:21 PM

the patch fixes the variability issue but not the linear increase of the duration

as you can see on one of the last attempt: https://logs.onap.org/onap-integration/weekly/onap_weekly_pod4_master/2021-06/11_22-55/stability/results_instantiation_10_24h//onaptests_bench.html

the duration is x2 on 24h, but it is better than SDC (x5 over 24h), it is tracked in another JIRA

Seshu Kumar Mudiganti June 16, 2021 at 1:40 PM

This issue seems to be resolved with the changes of the MariaDB patch on the oom.
Keeping it on hold for further confirmation.

Former user April 19, 2021 at 1:14 PM

Issue can also been tracked in daily:weekly CI as some steps timeout (SO looks stuck without timeout mechanism see https://lf-onap.atlassian.net/browse/SO-3584 and https://lf-onap.atlassian.net/browse/INT-1883

this issue seems to affect more a_la_carte_bpmn and more since Honoluu (DB backend change?)

I can be regularly observed in CI though pythonsdk_tests (see SO steps when duration is > 900s)

 

Seshu Kumar Mudiganti March 25, 2021 at 2:50 PM

we have been working on implementing the CSIT of SO flows with the same config as in the gating to see if we can check the problem in isolation. So far the macro flow of heat, Helm and ETSI have been implemented.
The issue seems to be not just with SO but the combination of it with other real component.
Still in process of cornering the issue.

Former user December 10, 2020 at 7:29 AM

the link corresponds to a snapshot after the installation and the first tests.

I attached the logs corresponding to 9 days of tests for bpmn-infra, catalog dB, openstack adapter..

we will reinstall this lab next sunday, if you need extra logs tell me before the reinstallation

Done

Details

Assignee

Reporter

Labels

Fix versions

Affects versions

Priority

Created December 9, 2020 at 2:58 PM
Updated January 19, 2022 at 2:59 PM
Resolved June 16, 2021 at 2:21 PM

Flag notifications