Hello devs,
according to [1], Apache has 180 job nodes total, shared for all Apache
projects. If we would migrate everything from travis to github actions,
we would use ~50 nodes for every created PR, PR update or integration.
Regarding travis [2], apache infra can't do anything about long travis
queue times, only if there are account related issues or similar
problems. So it appears the extremely long queue times (saw up to 20h in
some cases) which happen every few weeks are some kind of throttling and
no bug - but who knows.
Given this information, I believe we will have to basically keep doing
what we are already doing: split jobs between actions and travis. We
should also reconsider (again) if we really have to test everything in
every PR. PHP is the problem child right now since it has a 2.5h (!) job
in a 2x matrix with a relatively high probability that one of the two
axis randomly fails.
-> I don't think we can actually migrate to github actions from travis
opinions?
best regards,
michael
[1] https://infra.apache.org/github-actions-secrets.html
[2] https://issues.apache.org/jira/browse/INFRA-23257
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information about the NetBeans mailing lists, visit:
https://cwiki.apache.org/confluence/display/NETBEANS/Mailing+lists
