Project

General

Profile

Actions

action #30646

closed

[s390x][functional][fast][hard] Investigate why stress-tests fail starting with build 422.1

Added by nicksinger about 6 years ago. Updated about 6 years ago.

Status:
Resolved
Priority:
Urgent
Assignee:
Category:
Bugs in existing tests
Start date:
2018-01-22
Due date:
2018-02-13
% Done:

0%

Estimated time:
Difficulty:

Description

Description

All of them fail while preparing the SUT. However, the previous runs where all successful but without further investigation I struggle to find a suitable product issue I can blame.

Hypothesis:

  1. REJECTED okurz mentioned a hypothesis about an not properly escaped asterisk in the sed-command. A first quick look of mine in the previous runs couln't confirm this no, but to get rid of the warnings -> https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/4260
  2. ACCEPTED Did the network environment of this s390 instance change? -> might be, but found no evidence for this
  3. REJECTED test changed -> no test changes
  4. REJECTED Product bug -> crosschecked with different build/worker configurations - issue persists
  5. ACCEPTED s390p8 received an upgrade shortly before failing - there was a qemu update which might be the cause -> still we only could reproduce it sporadicly
  6. ACCEPTED s390p8 had a temporary hiccup - needs long-term statistics

Acceptance criteria:

Find out if this is a product bug (1) or test related (2).

(1): REJECTED Create a suitable bug report on bsc and try to relabel jobs pointing on this issue here -> no product bug, see H4

(2): REJECTED Fix the test so they can continue like before -> no test issue - see H3

Actions #1

Updated by okurz about 6 years ago

  • Subject changed from [s390x][functional] Investigate why stress-tests fail starting with build 422.1 to [s390x][functional][fast] Investigate why stress-tests fail starting with build 422.1
  • Due date set to 2018-01-30
  • Priority changed from Normal to Urgent
  • Target version set to Milestone 13

so you want to hide a potential critical product issue in our backlog somewhere late?

Actions #2

Updated by mgriessmeier about 6 years ago

  • Status changed from New to In Progress
  • Assignee set to mgriessmeier

I'll take a look

Actions #3

Updated by mgriessmeier about 6 years ago

so, it's failing since Build422.1 on o.s.d -> https://openqa.suse.de/tests/1405077
Build 421.1 was working fine on o.s.d -> https://openqa.suse.de/tests/1399411

manual tests conducted by cloning, based on latest upstream branch

manually reproducible with Build 422.1 -> opeth.suse.de/tests/13#step/boot_to_desktop/20
crosscheck with Build 421.1 results in the same error -> http://opeth/tests/14

restart of Build 421.1 on o.s.d also fails -> https://openqa.suse.de/tests/1412251

so I highly doubt that it's a product bug,
either the tests, backend or the environment changed. -> investigation ongoing

Actions #4

Updated by mgriessmeier about 6 years ago

  • Description updated (diff)
Actions #5

Updated by mgriessmeier about 6 years ago

soo.. update on this.

one day before the last good, I performed an upgrade on the s390 host which also installed a new qemu version
when I downgrade to the previous version it worked immediately on my machine. but when crosschecking with the new version again, it also worked...

Right now I'm doing some statistic runs with both versions

Actions #6

Updated by mgriessmeier about 6 years ago

hmmm all three stress tests are green again with Build 423.1 - https://openqa.suse.de/tests/1423558
qemu version at the moment is qemu-2.9.1-6.6.3.src.rpm, so the downgraded one...

forget about this... there is no s390 build in 423.1... I just took the wrong column
grabs coffee

Actions #7

Updated by riafarov about 6 years ago

  • Due date changed from 2018-01-30 to 2018-02-13
  • Target version changed from Milestone 13 to Milestone 14

Check in new build and then decide how to proceed.

Actions #8

Updated by okurz about 6 years ago

  • Subject changed from [s390x][functional][fast] Investigate why stress-tests fail starting with build 422.1 to [s390x][functional][fast][hard] Investigate why stress-tests fail starting with build 422.1

By now I assume it's a task for a "mob" :)

Actions #9

Updated by okurz about 6 years ago

  • Project changed from 46 to openQA Tests
  • Category set to Bugs in existing tests
  • Target version changed from Milestone 14 to Milestone 14
Actions #10

Updated by mgriessmeier about 6 years ago

  • Description updated (diff)
  • Status changed from In Progress to Resolved

working in Build439.1...

https://openqa.suse.de/tests/1447046
https://openqa.suse.de/tests/1447041
https://openqa.suse.de/tests/1447040

as discussed - adding Hypothesis and set to resolved
reopen if issue occurs again

Actions #11

Updated by mgriessmeier about 6 years ago

  • Description updated (diff)

updated ACs

Actions

Also available in: Atom PDF