Opened on 02/05/2019 at 11:52:32 PM
Last modified on 10/10/2019 at 11:11:46 AM
#7263 new defect
Gitlab runners timing out on adblockpluscore tests
| Reported by: | jsonesen | Assignee: | jsonesen |
|---|---|---|---|
| Priority: | Unknown | Milestone: | |
| Module: | Automation | Keywords: | |
| Cc: | mjethani, tlucas, sebastian, sergz | Blocked By: | |
| Blocking: | Platform: | Unknown / Cross platform | |
| Ready: | no | Confidential: | no |
| Tester: | Unknown | Verified working: | no |
| Review URL(s): |
https://gitlab.com/eyeo/adblockplus/adblockpluscore/merge_requests/20 |
||
Description
Environment
Gitlab CI
How to reproduce
- Fork eyeo project
- create a new branch
- npm install
- npm audit fix
- push changes to your upstream fork
...
Observed behaviour
Very inconsistent results, either timing out for over an hour or completing in minutes, or other times firefox driver errors propagate.
Expected behaviour
Test runner completes the job successfully, and in a timely fashion.
Notes
This seems to have something to do with the test runners either running out of memory or processing power, we have concluded this because test runners are autoscaled meaning it is sort of up to the ci what type of image your runner executes on, and sometimes (most of the time for me) they seem to be to weak.
What to change
I already have a patch which fixes this by breaking the job into multiple npm test calls; one for each test script in the test dir. So far serializing the tests seemed to help immensely, but I would like to find a slightly more elegant solution than manually telling the yaml each test file to run.
