You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The PRs #1468 and #1481 had some issues with the CI failing because of memory allocation errorsFATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory. Looking into it, it seems there are some known issues with jest related to this:
I did some comparisons with different solutions using --runInBand --logHeapUsage but none of the changes had any clear impact. Some issues also mention that there are problems with --runInBand so could also be related to that. In the end I disabled coverage for integration tests and added the main suggestions from the linked PRs (which is adding the --expose-gc --no-compilation-cache node flags) in 80fa81a, which seems to have resolved the issue.
This issue is mostly a reminder to perhaps look more extensively into this in the future, and also in case this issue should occur again.
The text was updated successfully, but these errors were encountered:
I additionally had to increase the integration test timeout to 20 minutes as 15 minutes was no longer sufficient.
I wonder if we can somehow split our set of integration tests into several different actions. That way they would be resolved faster, allowing more of them to be done in parallel, and would also reduce the memory consumption issue as that is scoped per action.
The PRs #1468 and #1481 had some issues with the CI failing because of memory allocation errors
FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory
. Looking into it, it seems there are some known issues with jest related to this:I did some comparisons with different solutions using
--runInBand --logHeapUsage
but none of the changes had any clear impact. Some issues also mention that there are problems with--runInBand
so could also be related to that. In the end I disabled coverage for integration tests and added the main suggestions from the linked PRs (which is adding the--expose-gc --no-compilation-cache
node flags) in 80fa81a, which seems to have resolved the issue.This issue is mostly a reminder to perhaps look more extensively into this in the future, and also in case this issue should occur again.
The text was updated successfully, but these errors were encountered: