Replace tot with snowflake IDs

[krzyzacy]

ref #9593 (comment)

/area prow
cc @BenTheElder @cjwagner @spiffxp

[spiffxp]

/area gubernator
Might have some impact here?

/area kettle
We would definitely want to revisit kettle's impl

/area testgrid
I don't know if this would impact testgrid at all, maybe during transition?

[BenTheElder]

testgrid should be unimpacted unless we decide to go back, going back would be problematic. gubernator otoh reads latest-build.txt and would be fine.

[ixdy]

/assign @fejta @BenTheElder
cc @Katharine to fix gubernator
cc @michelle192837 to fix testgrid and kettle

[BenTheElder]

I think testgrid already works fine? not sure about kettle cc @spiffxp
Gubernator will need fixing or we'll need spyglass to be ready though

[michelle192837]

TestGrid's fine in its current implementation as long as the IDs monotonically increase. It'll need tweaks if not.

[BenTheElder]

they're monotonic, just with gaps. testgrid is already handling some timestamp IDs so we should be good 👍
http://testgrid.k8s.io/conformance-cloud-provider-openstack#periodic-master

[Katharine]

Made the required gubernator change in #9911. It's slow.

Gubernator already has fallback logic in case latest-build.txt doesn't exist, so it might instead be preferable to just ensure there is no latest-build.txt?

[Katharine]

Gubernator changes are live, so that blocker is gone.

[BenTheElder]

so probably just check kettle and then we can look at deleting the service

[Katharine]

Looked into this briefly, we also need to set sequential: false as appropriate in kettle/buckets.yaml before we kill tot.

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[Katharine]

/remove-lifecycle stale

There are no known (to me) blockers on this. We just need to pick a time we don't expect to matter an awful lot if things are on fire. After we work through the merge backlog at the beginning of the 1.15 cycle maybe?

We'll also have to loop in contribex and maybe kubernetes-dev because the numbers will be large and ugly and that's a change.

[spiffxp]

I'm in favor of trying this out, and yes we should give kubernetes-dev@ a heads up. My only ask is that we wait until v1.14 goes out the door.

[Katharine]

After 1.14 ships is my plan.

[stevekuznetsov]

/woof

[k8s-ci-robot]

@stevekuznetsov:

In response to this:

/woof

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[spiffxp]

/pony snowflake

[k8s-ci-robot]

@spiffxp:

In response to this:

/pony snowflake

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[Katharine]

We did it!

/pony cheering
/close

[k8s-ci-robot]

@Katharine: Closing this issue.

In response to this:

We did it!

/pony cheering
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[k8s-ci-robot]

@Katharine:

In response to this:

We did it!

/pony cheering
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[krzysied]

Hi folks!
Currently we are having issue with perfdash - it doesn't work. kubernetes/perf-tests#469
Assuming that you made changes to the other tools, can someone give hints on how to work with the new run id format? We are using k8s.io/contrib/test-utils/utils/utils.go for accessing the run logs.