Shift productivity infrastructure to a community managed environment about community HOT 20 CLOSED

Looks like the Mako folks are giving up: google/mako#2 (comment)

from community.

Any community alternatives to Mako?

from community.

@bbrowning What do the Openshift perf folks use for dashboarding?

from community.

@mattmoor I'm not sure we have any tools which are exposed publicly, but I'll ask around to see what they use internally to see if that's something we could use in the community.

from community.

@bsnchan does this issue cover just the performance stuff or do you see it covering all tooling used by knative? e.g. github management (e.g prow), release building, PR testing, etc...

It would be great if we could expand it. Start by enumerating all of the bits that (today) require a specific cloud provider, and then work on 1) tweaking the code/scripts to support multiple providers, and 2) look for volunteers to spread the cost and hosting responsibilities.

from community.

It covers everything, I was just noting that Mako was particularly closed.

from community.

Mako is now going in to maintenance mode; so if there are other alternatives people know of and like; might be a good time to review those? Mako will still be supported but no new features; and PII remains an issue for getting a Mako version that can be accessed by all.

from community.

Since things have changed a little bit since the last time this was looked at (github actions for one), I'm going to start the list of tools that I'm aware of so that we can evaluate options. @duglin @mattmoor @n3wscott what others are missing from here? Also @bbrowning were you able to find anything out?

• testgrid
• mako
• prow
• release jobs

from community.

more than just prow, there are several jobs that run in prow that do work for us. So finding an alternative for those jobs will be needed and likely require some dev work.

from community.

@n3wscott where should I look for that list of jobs? Thanks :)

from community.

I think here @vaikas https://github.com/GoogleCloudPlatform/oss-test-infra/tree/master/prow/knative

from community.

Looking through that list and the generated config, I see:

• Test log storage and viewing. (https://github.com/kubernetes/test-infra/tree/master/prow/plank, https://github.com/kubernetes/test-infra/tree/master/prow/spyglass + Google Cloud Storage)
• Slack reporting
• Automatic merging with testing. (https://github.com/kubernetes/test-infra/tree/master/prow/tide)
• Label management for non-writers (https://github.com/GoogleCloudPlatform/oss-test-infra/blob/master/prow/knative/plugins.yaml#L99 -- /area et al)
• Test/retest for non-writers
• Cherrypicker (https://github.com/kubernetes/test-infra/tree/master/prow/external-plugins/cherrypicker)

from community.

I'm wary of moving more things to GH actions given our ongoing issues with queuing issues, and the lack of visibility.
Prow has also seen tons of improvements as of late, so things are getting better on that front.

from community.

@evankanderson suggested using the Github actions with runners into prow cluster.

from community.

See also https://docs.google.com/document/d/1LwkzXVDJYLmiVElCEiJ1ITSr4TGkHfipe8ABK_7IxCQ/edit?hl=en#heading=h.ky4f33l2ky29

from community.

Steering doesn't consider this to be critical for 1.0, our productivity group may be spread too thin at the moment so we may not pick this up until post 1.0 (unless someone raises their hand to help with this in the meantime).

from community.

cc @chizhg just presented a roadmap document during the ToC

https://docs.google.com/document/d/1U6wybYjCslEhsKOiennrwhoKZmSVnd1NGbFI8gnnnOg/edit#

from community.

related/dupe
#786

from community.

/close

Dup of #786

from community.

@evankanderson: Closing this issue.

from community.