New issue
Advanced search Search tips

Issue 654623 link

Starred by 3 users

Issue metadata

Status: Fixed
Owner:
Closed: Nov 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 3
Type: Bug



Sign in to add a comment

Milo doesn't show pending or current builds

Project Member Reported by martiniss@chromium.org, Oct 11 2016

Issue description

Comment 1 by d...@chromium.org, Oct 11 2016

Cc: hinoka@chromium.org
Owner: ----
Let's not implicitly assign ownership.

I also note that even among complete builds it's lagging. Latest complete build is #26577, latest in Milo is #26586. The Pub/Sub subscription does not have a backlog.

Searching through logs:
https://pantheon.corp.google.com/logs/viewer?project=luci-milo&key1=default&minLogLevel=0&expandAll=false&resource=appengine.googleapis.com&logName=projects%2Fluci-milo%2Flogs%2Fappengine.googleapis.com%252Frequest_log&advancedFilter=metadata.serviceName%3D%22appengine.googleapis.com%22%0Alog%3D%22appengine.googleapis.com%2Frequest_log%22%0AprotoPayload.resource%3D%22%2Fpubsub%2Fbuildbot%22%0A%22Checking%20for%20build%20chromium.perf%2FWin%20x64%20Builder%22&lastVisibleTimestampNanos=1476150933727961000

The last message pushed from the master is at 6:55 PM (it is 7:03).

My guess: the Twisted process is at 100% CPU and is overloaded, and is therefore not pushing out updates.

Possible solution that come to mind:
1) Rewrite Pub/Sub push to work around this sort of problem. This might not work b/c BuildBot is the core problem here.
2) Remove BuildBot logging from chromium.perf in favor of LogDog. Master load will probably drop to 1% :)

Comment 2 by d...@chromium.org, Oct 11 2016

CPU for this master is definitely capped out (149% on "top").

Comment 3 by d...@chromium.org, Oct 11 2016

Also no exceptions in the master's log.

Comment 4 by hinoka@chromium.org, Oct 11 2016

Owner: hinoka@chromium.org
Status: Assigned (was: Unconfirmed)
I'll take this, I found a few bugs in the pipeline.
Project Member

Comment 5 by bugdroid1@chromium.org, Oct 11 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build.git/+/fb0560def63fcd666d8ef9ece802d88622d200fc

commit fb0560def63fcd666d8ef9ece802d88622d200fc
Author: hinoka <hinoka@google.com>
Date: Tue Oct 11 21:43:12 2016

pubsub: pending fix

1. Rename misspelled "pendingBuildStatues" to "pendingBuildStates"
2. Yieldind a deferred list returns a list of tupes with success, we only want
to return the data.

BUG= 654623 

Review-Url: https://codereview.chromium.org/2406133002

[modify] https://crrev.com/fb0560def63fcd666d8ef9ece802d88622d200fc/scripts/master/pubsub_json_status_push.py

Project Member

Comment 6 by bugdroid1@chromium.org, Oct 11 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build.git/+/fb0560def63fcd666d8ef9ece802d88622d200fc

commit fb0560def63fcd666d8ef9ece802d88622d200fc
Author: hinoka <hinoka@google.com>
Date: Tue Oct 11 21:43:12 2016

pubsub: pending fix

1. Rename misspelled "pendingBuildStatues" to "pendingBuildStates"
2. Yieldind a deferred list returns a list of tupes with success, we only want
to return the data.

BUG= 654623 

Review-Url: https://codereview.chromium.org/2406133002

[modify] https://crrev.com/fb0560def63fcd666d8ef9ece802d88622d200fc/scripts/master/pubsub_json_status_push.py

Project Member

Comment 7 by bugdroid1@chromium.org, Oct 12 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/external/github.com/luci/luci-go.git/+/ab30d82eb01708c79fec5db8833523af53d8169e

commit ab30d82eb01708c79fec5db8833523af53d8169e
Author: hinoka <hinoka@google.com>
Date: Wed Oct 12 02:18:48 2016

Milo: Display pending builds

This fixes some bugs (naming mismatches) with the receiving end of pubsub, and
also adds pending build data into a builder view.

The rename was done here:
https://codereview.chromium.org/2406133002/

BUG= 654623 

Review-Url: https://codereview.chromium.org/2409953002

[modify] https://crrev.com/ab30d82eb01708c79fec5db8833523af53d8169e/milo/appengine/buildbot/builder.go
[modify] https://crrev.com/ab30d82eb01708c79fec5db8833523af53d8169e/milo/appengine/buildbot/structs.go

Project Member

Comment 8 by bugdroid1@chromium.org, Oct 12 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/external/github.com/luci/luci-go.git/+/ab30d82eb01708c79fec5db8833523af53d8169e

commit ab30d82eb01708c79fec5db8833523af53d8169e
Author: hinoka <hinoka@google.com>
Date: Wed Oct 12 02:18:48 2016

Milo: Display pending builds

This fixes some bugs (naming mismatches) with the receiving end of pubsub, and
also adds pending build data into a builder view.

The rename was done here:
https://codereview.chromium.org/2406133002/

BUG= 654623 

Review-Url: https://codereview.chromium.org/2409953002

[modify] https://crrev.com/ab30d82eb01708c79fec5db8833523af53d8169e/milo/appengine/buildbot/builder.go
[modify] https://crrev.com/ab30d82eb01708c79fec5db8833523af53d8169e/milo/appengine/buildbot/structs.go

Project Member

Comment 9 by bugdroid1@chromium.org, Oct 12 2016

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infradata/master-manager.git/+/f5224160c37436254be092b669d93f766eb8129c

commit f5224160c37436254be092b669d93f766eb8129c
Author: hinoka <hinoka@google.com>
Date: Wed Oct 12 05:39:07 2016

Summary: Milo doesn't show pending or current builds (was: Milo doesn't show pending builds)
Reusing this bug for a similar issue...
Pending builds show up (woooo)

Only for recently restarted masters though
https://luci-milo.appspot.com/buildbot/chromium.fyi/CrWinClang%28dbg%29/
Labels: W-1017
Ping!

Any progress on current builds?

Pending builds now show up, but not current builds.
Project Member

Comment 15 by bugdroid1@chromium.org, Oct 26 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/external/github.com/luci/luci-go.git/+/a94e9e33c00d7ecc2acd4a8489732f026c768a35

commit a94e9e33c00d7ecc2acd4a8489732f026c768a35
Author: hinoka <hinoka@google.com>
Date: Wed Oct 26 21:39:47 2016

Milo: Clean up builds that are no longer current.

Sometimes when masters restart, they don't flush builds, so that current builds
continue to linger in Milo despite being expired (and finished).

This uses the signal that if a build exists in Milo but isn't in the master
JSON, then it should get marked as expired.

BUG= 654623 

Review-Url: https://codereview.chromium.org/2406163003

[modify] https://crrev.com/a94e9e33c00d7ecc2acd4a8489732f026c768a35/milo/appengine/buildbot/pubsub.go
[modify] https://crrev.com/a94e9e33c00d7ecc2acd4a8489732f026c768a35/milo/appengine/buildbot/pubsub_test.go

Project Member

Comment 16 by bugdroid1@chromium.org, Oct 26 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/external/github.com/luci/luci-go.git/+/a94e9e33c00d7ecc2acd4a8489732f026c768a35

commit a94e9e33c00d7ecc2acd4a8489732f026c768a35
Author: hinoka <hinoka@google.com>
Date: Wed Oct 26 21:39:47 2016

Milo: Clean up builds that are no longer current.

Sometimes when masters restart, they don't flush builds, so that current builds
continue to linger in Milo despite being expired (and finished).

This uses the signal that if a build exists in Milo but isn't in the master
JSON, then it should get marked as expired.

BUG= 654623 

Review-Url: https://codereview.chromium.org/2406163003

[modify] https://crrev.com/a94e9e33c00d7ecc2acd4a8489732f026c768a35/milo/appengine/buildbot/pubsub.go
[modify] https://crrev.com/a94e9e33c00d7ecc2acd4a8489732f026c768a35/milo/appengine/buildbot/pubsub_test.go

Almost, these are the CLs left:

Bug fix with expiring builds: https://codereview.chromium.org/2446273005/
Add ?limit=... https://codereview.chromium.org/2328733002/
Actually add it to the view: https://codereview.chromium.org/2321243002
Project Member

Comment 18 by bugdroid1@chromium.org, Oct 27 2016

The following revision refers to this bug:
  https://chromium.googlesource.com/external/github.com/luci/luci-go.git/+/b0a9fe11208861902c7767fbb3eb7adc98bff727

commit b0a9fe11208861902c7767fbb3eb7adc98bff727
Author: hinoka <hinoka@google.com>
Date: Thu Oct 27 23:38:16 2016

Milo: Pubsub pending build fixes.

This fixes the large pending build (and master json size) issue by stripping out
all files, comments, and property information from pending builds (they aren't
used or displayed anywhere anyways.)

This shrunk chromium.fyi's compressed json size from 2MB down to 60KB

This also cleans up some debug statements.

BUG= 654623 

Review-Url: https://codereview.chromium.org/2450343003

[modify] https://crrev.com/b0a9fe11208861902c7767fbb3eb7adc98bff727/milo/appengine/buildbot/pubsub.go

Sign in to add a comment