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
{{ message }}
Repository navigation
Check that a semver-major PR has at least a CITGM CI run #141
Could we include semver-minor in this? In general, more PRs should get CitGM runs than is currently the case. I've started running it for any change I make that touches internals in a significant way, even if it seems completely harmless. The CitGM job doesn't even take much longer to finish than the main CI.
@apapirovski Well now come to think of it, maybe we should just create a label "citgm-needed"? The rule would be "semvor-major || citgm-needed".
I thought the reason why we don't run CITGM that often was because it's slow or we don't have enough resources (e.g. machines), but that might have changed or was never true... cc @nodejs/build @nodejs/citgm
I thought the reason why we don't run CITGM that often was because it's slow or we don't have enough resources (e.g. machines), but that might have changed or was never true.
The biggest issue with CitGM is that it needs triage, modules fail all the time, usually unrelated to node core. There are also lots of modules with flaky tests. I don't think the runs have ever taken much more than an hour.
On a side note, we should check that the latest CI runs should be reasonably fresh (at least not older than a week?)
A week seems reasonable to me, after that the runs get removed from Jenkins anyway, so aren't really trustworthy.
This issue is stale because it has been open many days with no activity. It will be closed soon unless the stale label is removed or a comment is made.
Activity
apapirovski commentedon Jan 17, 2018
Could we include semver-minor in this? In general, more PRs should get CitGM runs than is currently the case. I've started running it for any change I make that touches internals in a significant way, even if it seems completely harmless. The CitGM job doesn't even take much longer to finish than the main CI.
joyeecheung commentedon Jan 17, 2018
@apapirovski Well now come to think of it, maybe we should just create a label "citgm-needed"? The rule would be "semvor-major || citgm-needed".
I thought the reason why we don't run CITGM that often was because it's slow or we don't have enough resources (e.g. machines), but that might have changed or was never true... cc @nodejs/build @nodejs/citgm
apapirovski commentedon Jan 17, 2018
FWIW the last successful CitGM job took 1hr16min within most results being available after the first 30-40 minutes.
richardlau commentedon Jan 17, 2018
Also there are modules tested by CitGM that are known to fail in master: nodejs/citgm#517
gibfahn commentedon Jan 17, 2018
The biggest issue with CitGM is that it needs triage, modules fail all the time, usually unrelated to node core. There are also lots of modules with flaky tests. I don't think the runs have ever taken much more than an hour.
A week seems reasonable to me, after that the runs get removed from Jenkins anyway, so aren't really trustworthy.
github-actions commentedon Aug 17, 2020
This issue is stale because it has been open many days with no activity. It will be closed soon unless the stale label is removed or a comment is made.