-
Notifications
You must be signed in to change notification settings - Fork 17.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
x/build: reverse builder config uniformity #21260
Comments
I will start referencing bugs from here. Here is why the Solaris/illumos builders are not ideal: #15581 |
Change https://golang.org/cl/52633 mentions this issue: |
Change https://golang.org/cl/52650 mentions this issue: |
I deployed CL 52633 to see which hosts are using the old
Updating the Scaleway images can happen in https://go-review.googlesource.com/c/52293 A bunch of the others are done in https://go-review.googlesource.com/c/52650 (once ppc64 and arm5-spacemonkey are updated) Updating Mac images is super hard and tedious and SLOW and error prone. I think we'd be better of keeping the mapping in the coordinator and buildlet for a bit. We can hide it from flag listing if helpful to minimize doc clutter. MIPS seems to be updateable easily enough. I also want to figure out a plan for #21191 (speaking of cleanup). |
…cy config We want to kill off the --reverse vs --reverse-host code. This will point out any clients still using the old way. There might not be any. Updates golang/go#21260 Change-Id: Ic4b283b02891655ee43f2b7db76088df3f12065a Reviewed-on: https://go-review.googlesource.com/52633 Reviewed-by: Sarah Adams <shadams@google.com>
Updates golang/go#21260 Change-Id: Ic4d96ef3984017944c5a9af087ea7010e5193f32 Reviewed-on: https://go-review.googlesource.com/52650 Reviewed-by: Jessica Frazelle <jessfraz@google.com>
Latest update on which hosts are still using the legacy ways:
Basically no change since Aug 1 2017, except that we no longer OS X 10.8 and the MIPS builders disappeared. The one fix seems to be that host-linux-ppc64le-osu and host-linux-ppc64-osu are no longer doing the legacy registration. Also, macOS 10.14 is at least using the new way. And the Mac VMware cluster is easier to work with now since it was upgraded, so we could fix those old versions at least. |
Two of those builders are gone now, so remaining is:
|
Change https://golang.org/cl/200757 mentions this issue: |
One of the little ppc64le VMs was lost in the upgrade, and we got one new beefy one. Updates golang/go#34830 Updates golang/go#21189 Updates golang/go#21260 Change-Id: I4488d106c5c146c6e55d2d3528913f5507b9d57b Reviewed-on: https://go-review.googlesource.com/c/build/+/200757 Reviewed-by: Bryan C. Mills <bcmills@google.com>
Change https://golang.org/cl/200820 mentions this issue: |
Add details of 3 new big VMs: POWER8 (le), POWER9 (le), and POWER8 (be). Still not configured, but we have them provisioned. Updates golang/go#34830 Updates golang/go#21189 Updates golang/go#21260 Change-Id: If648e9966485062cda8131caf2ac5f87ab839fc5 Reviewed-on: https://go-review.googlesource.com/c/build/+/200820 Reviewed-by: Bryan C. Mills <bcmills@google.com>
Change https://golang.org/cl/201637 mentions this issue: |
Fixes golang/go#21189 Updates golang/go#21260 Updates golang/go#34830 (ppc64 big endian remains) Change-Id: I7ff38edb519bf4f34cd05bb795b6a41f72b10640 Reviewed-on: https://go-review.googlesource.com/c/build/+/201637 Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com> Reviewed-by: Bryan C. Mills <bcmills@google.com>
Change https://golang.org/cl/203886 mentions this issue: |
Collaboration with @tiborvass at Docker who got Docker running on big-endian PPC64. Go for ppc64 doesn't support cgo or external linking, so runc doesn't work, but a new OCI-compliant runc implementation written in C (https://github.com/containers/crun) means we can run Docker after all. See NOTES & build-*.sh Then add a Dockerfile & associated cleanup in buildlet & stage0 to use rundockerbuildlet. Once done, might help with golang/go#35188, golang/go#32613, etc. Fixes golang/go#34830 Updates golang/go#21260 Change-Id: I43d7afa1d58bbdfa16e3c57670bc41f1d1932d80 Reviewed-on: https://go-review.googlesource.com/c/build/+/203886 Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Down to:
|
Just double checked, and yes:
|
And I just fixed the darwin ones. |
Change https://golang.org/cl/205357 mentions this issue: |
…for Macs Updates golang/go#21260 Change-Id: I4e2c61117cfebffd20ab4acd3ee736172c95d153 Reviewed-on: https://go-review.googlesource.com/c/build/+/205357 Reviewed-by: Bryan C. Mills <bcmills@google.com>
Change https://golang.org/cl/205603 mentions this issue: |
Change https://golang.org/cl/205608 mentions this issue: |
…erse flag Updates golang/go#21260 (no more buildlets using the --reverse flag) Updates golang/go#33574 (linux-arm kernel+userspace updated) Change-Id: I7455f6fa3e851f1f9f81d6f1eb487ef7e4bea55b Reviewed-on: https://go-review.googlesource.com/c/build/+/205603 Reviewed-by: Bryan C. Mills <bcmills@google.com>
One of the little ppc64le VMs was lost in the upgrade, and we got one new beefy one. Updates golang/go#34830 Updates golang/go#21189 Updates golang/go#21260 Change-Id: I4488d106c5c146c6e55d2d3528913f5507b9d57b Reviewed-on: https://go-review.googlesource.com/c/build/+/200757 Reviewed-by: Bryan C. Mills <bcmills@google.com>
Add details of 3 new big VMs: POWER8 (le), POWER9 (le), and POWER8 (be). Still not configured, but we have them provisioned. Updates golang/go#34830 Updates golang/go#21189 Updates golang/go#21260 Change-Id: If648e9966485062cda8131caf2ac5f87ab839fc5 Reviewed-on: https://go-review.googlesource.com/c/build/+/200820 Reviewed-by: Bryan C. Mills <bcmills@google.com>
Fixes golang/go#21189 Updates golang/go#21260 Updates golang/go#34830 (ppc64 big endian remains) Change-Id: I7ff38edb519bf4f34cd05bb795b6a41f72b10640 Reviewed-on: https://go-review.googlesource.com/c/build/+/201637 Reviewed-by: Carlos Eduardo Seo <cseo@linux.vnet.ibm.com> Reviewed-by: Bryan C. Mills <bcmills@google.com>
Collaboration with @tiborvass at Docker who got Docker running on big-endian PPC64. Go for ppc64 doesn't support cgo or external linking, so runc doesn't work, but a new OCI-compliant runc implementation written in C (https://github.com/containers/crun) means we can run Docker after all. See NOTES & build-*.sh Then add a Dockerfile & associated cleanup in buildlet & stage0 to use rundockerbuildlet. Once done, might help with golang/go#35188, golang/go#32613, etc. Fixes golang/go#34830 Updates golang/go#21260 Change-Id: I43d7afa1d58bbdfa16e3c57670bc41f1d1932d80 Reviewed-on: https://go-review.googlesource.com/c/build/+/203886 Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
…for Macs Updates golang/go#21260 Change-Id: I4e2c61117cfebffd20ab4acd3ee736172c95d153 Reviewed-on: https://go-review.googlesource.com/c/build/+/205357 Reviewed-by: Bryan C. Mills <bcmills@google.com>
…erse flag Updates golang/go#21260 (no more buildlets using the --reverse flag) Updates golang/go#33574 (linux-arm kernel+userspace updated) Change-Id: I7455f6fa3e851f1f9f81d6f1eb487ef7e4bea55b Reviewed-on: https://go-review.googlesource.com/c/build/+/205603 Reviewed-by: Bryan C. Mills <bcmills@google.com>
All the buildlets have been updated to use --reverse-type with a host type instead of a builder type. Fixes golang/go#21260 Change-Id: I1264261f099c3686fe01455949486f523b94c6de Reviewed-on: https://go-review.googlesource.com/c/build/+/205608 Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org> TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Bryan C. Mills <bcmills@google.com>
Onboarding overhead is very high ATM for builder project.
I'd like to discuss a config uniformity overhaul for the reverse buildlets.
We have been migrating various reverse buildlet Dockerfiles / configurations over to various different stages of "new" piecemeal b/c lack of headcount. eg. uses X version of rundockerbuildlet, uses rundockerbuildlet, are builds hermetic?, uses
--reverse
(legacy) vs.--reverse-type
(latest) flag tocmd/buildlet/buildlet.go
I'd like us document the ideal/latest blessed reverse buildlet config, and push to bring all reverse buildlet configurations up to this config scheme.
I'd also like to discuss the potential for a plan to keep reverse buildlet configurations at
latest
config scheme in the future.The text was updated successfully, but these errors were encountered: