LUCI TryBots now required for submission

[Michael Pratt]

If you’re already using LUCI TryBots by voting Commit-Queue+1, you can stop reading here.

So far it’s been possible to use either Commit-Queue+1 (“Run LUCI TryBots”) or Run-TryBot+1 (“Run legacy TryBots”) to get a passing result and satisfy the “TryBot-Pass” submit requirement, because by default they provide equivalent signal.

The LUCI TryBots have been available since August, 2023, and we believe they’re ready to become the default pre-submit testing method for the Go project. Thus, as of today, we have applied a configuration change to make it a submit requirement on every branch to have a passing LUCI TryBot result.

It is still possible to trigger legacy SlowBots if there is a specific builder you'd like to run which isn’t yet available as a LUCI builder [1]. If legacy TryBots are requested, submission will require that they pass (to prevent auto-submit from submitting with failing legacy TryBots). This doesn’t replace the need to also have a passing LUCI TryBot result.

If this causes unexpected problems for your workflow, please reply to this thread or file an issue with an “x/build:” prefix.

Make sure you have updated to a recent version of git-codereview (`go install golang.org/x/review/git-codereview@latest`) to have `git codereview mail -trybot` target LUCI TryBots.

[1] If a builder is available on LUCI, see the LUCI SlowBots instructions for triggering SlowBots with LUCI.

Thanks,
Michael

[qiulaidongfeng]

I'm not sure if I need to submit an issue, but I found Commit-Queue+1 takes about 20 minutes at a time. Run-TryBot+1 takes about 14 minutes per time. What I have observed is that linux-amd64-race is significantly slower at Commit-Queue+1 than Run-TryBot+1. Is this intentional?

[Michael Pratt]

On Sat, Feb 17, 2024 at 5:38 PM 'qiulaidongfeng' via golang-dev <golan...@googlegroups.com> wrote:

I'm not sure if I need to submit an issue, but I found Commit-Queue+1 takes about 20 minutes at a time. Run-TryBot+1 takes about 14 minutes per time. What I have observed is that linux-amd64-race is significantly slower at Commit-Queue+1 than Run-TryBot+1. Is this intentional?

The test sharding approach that LUCI builders is a bit different from how we did sharding on the old infrastructure. The new sharding is often unbalanced (the race builder tends to have the slowest shard take ~12min, and fastest take ~4min), which leads to longer total latency. We'd like to improve this, but didn't have a bug, so we're filed https://go.dev/issue/65814 to track this.

 

On Saturday, February 17, 2024 at 4:13:33 AM UTC+8 Michael Pratt wrote:

If you’re already using LUCI TryBots by voting Commit-Queue+1, you can stop reading here.

So far it’s been possible to use either Commit-Queue+1 (“Run LUCI TryBots”) or Run-TryBot+1 (“Run legacy TryBots”) to get a passing result and satisfy the “TryBot-Pass” submit requirement, because by default they provide equivalent signal.

The LUCI TryBots have been available since August, 2023, and we believe they’re ready to become the default pre-submit testing method for the Go project. Thus, as of today, we have applied a configuration change to make it a submit requirement on every branch to have a passing LUCI TryBot result.

It is still possible to trigger legacy SlowBots if there is a specific builder you'd like to run which isn’t yet available as a LUCI builder [1]. If legacy TryBots are requested, submission will require that they pass (to prevent auto-submit from submitting with failing legacy TryBots). This doesn’t replace the need to also have a passing LUCI TryBot result.

If this causes unexpected problems for your workflow, please reply to this thread or file an issue with an “x/build:” prefix.

Make sure you have updated to a recent version of git-codereview (`go install golang.org/x/review/git-codereview@latest`) to have `git codereview mail -trybot` target LUCI TryBots.

[1] If a builder is available on LUCI, see the LUCI SlowBots instructions for triggering SlowBots with LUCI.

Thanks,
Michael

--
You received this message because you are subscribed to the Google Groups "golang-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/golang-dev/7e0a2b9e-f4a4-4d31-9749-a4441230e9c6n%40googlegroups.com.

[qiulaidongfeng]

How to add cq-included-trybots: luci.golang.try:gotip-linux-386-longtest in CL contributed by github PR?

I want to do so at https://go-review.googlesource.com/c/go/+/572195, Because similar CL has OOM in linux-386-longtest before, but LUCI-TryBot using Commit-Queue+1 by default has not been tested, so it has been revert.

Edit directly https://github.com/golang/go/pull/66359#issue-2190435437 and Change-Id will have empty line.

[Michael Pratt]

On Sun, Mar 17, 2024 at 11:41 PM 'qiulaidongfeng' via golang-dev <golan...@googlegroups.com> wrote:

How to add cq-included-trybots: luci.golang.try:gotip-linux-386-longtest in CL contributed by github PR?

I want to do so at https://go-review.googlesource.com/c/go/+/572195, Because similar CL has OOM in linux-386-longtest before, but LUCI-TryBot using Commit-Queue+1 by default has not been tested, so it has been revert.

Edit directly https://github.com/golang/go/pull/66359#issue-2190435437 and Change-Id will have empty line.

Did you actually try this and have it break? https://go.dev/cl/564837 is intended to ensure that a Cq-Include-Trybots line in the GitHub description is grouped with the Change-Id line.

Also, note that it is "Cq-Include-Trybots:". Your message says "cq-included-trybots:" ("included" vs "include"). I'm not sure if casing matters as well.

 

在2024年2月21日星期三 UTC+8 06:06:17<Michael Pratt> 写道：

On Sat, Feb 17, 2024 at 5:38 PM 'qiulaidongfeng' via golang-dev <golan...@googlegroups.com> wrote:

I'm not sure if I need to submit an issue, but I found Commit-Queue+1 takes about 20 minutes at a time. Run-TryBot+1 takes about 14 minutes per time. What I have observed is that linux-amd64-race is significantly slower at Commit-Queue+1 than Run-TryBot+1. Is this intentional?

The test sharding approach that LUCI builders is a bit different from how we did sharding on the old infrastructure. The new sharding is often unbalanced (the race builder tends to have the slowest shard take ~12min, and fastest take ~4min), which leads to longer total latency. We'd like to improve this, but didn't have a bug, so we're filed https://go.dev/issue/65814 to track this.

 

On Saturday, February 17, 2024 at 4:13:33 AM UTC+8 Michael Pratt wrote:

If you’re already using LUCI TryBots by voting Commit-Queue+1, you can stop reading here.

So far it’s been possible to use either Commit-Queue+1 (“Run LUCI TryBots”) or Run-TryBot+1 (“Run legacy TryBots”) to get a passing result and satisfy the “TryBot-Pass” submit requirement, because by default they provide equivalent signal.

The LUCI TryBots have been available since August, 2023, and we believe they’re ready to become the default pre-submit testing method for the Go project. Thus, as of today, we have applied a configuration change to make it a submit requirement on every branch to have a passing LUCI TryBot result.

It is still possible to trigger legacy SlowBots if there is a specific builder you'd like to run which isn’t yet available as a LUCI builder [1]. If legacy TryBots are requested, submission will require that they pass (to prevent auto-submit from submitting with failing legacy TryBots). This doesn’t replace the need to also have a passing LUCI TryBot result.

If this causes unexpected problems for your workflow, please reply to this thread or file an issue with an “x/build:” prefix.

Make sure you have updated to a recent version of git-codereview (`go install golang.org/x/review/git-codereview@latest`) to have `git codereview mail -trybot` target LUCI TryBots.

[1] If a builder is available on LUCI, see the LUCI SlowBots instructions for triggering SlowBots with LUCI.

Thanks,
Michael

--
You received this message because you are subscribed to the Google Groups "golang-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/golang-dev/7e0a2b9e-f4a4-4d31-9749-a4441230e9c6n%40googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "golang-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to golang-dev+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/golang-dev/c948d409-08ca-4125-9638-19002438bc41n%40googlegroups.com.

[qiulaidongfeng]

Thanks reply.

I'm sorry I misunderstood 

That is, it must be right next to the Change-Id line without empty line. in https://go.dev/wiki/SlowBots#using-slowbots

Because I didn't know before that CL contributed via github PR can do that there is no blank line between the submission information and the Change-Id line. I accidentally misinterpreted that to mean that modifying the commit information in github PR to add  Cq-Include-Trybots: luci.golang.try:gotip-linux-386-longtest   is not ok.

[Than McIntosh]

For what it's worth, the "no blank lines" thing got me a couple of times as well (couldn't figure out what it was that I wasn't doing correctly).

Cheers, Than

To view this discussion on the web visit https://groups.google.com/d/msgid/golang-dev/c373df6b-8624-4e83-913b-a7d6208a7190n%40googlegroups.com.