New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 710239 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner:
Closed: Aug 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: ----

Blocking:
issue 710128



Sign in to add a comment

Purple Bot on chromium.perf: mac_pro_perf_bisect

Project Member Reported by simonhatch@chromium.org, Apr 10 2017

Issue description

Link to buildbot status page:
https://build.chromium.org/p/tryserver.chromium.perf/buildslaves/build196-b4


INFO:root:Running: ('git', 'clone', '-q', 'https://chromium.googlesource.com/chromium/tools/build.git', '/b/rr/tmpxXDdDK/rw/checkout')
fatal: write error: No space left on device
fatal: index-pack failed
Traceback (most recent call last):
  File "/b/.remote_run_cipd/recipes.py", line 747, in <module>
    ret = main()
  File "/b/.remote_run_cipd/recipes.py", line 663, in main
    return _real_main(args, op_args)
  File "/b/.remote_run_cipd/recipes.py", line 684, in _real_main
    return remote(args)
  File "/b/.remote_run_cipd/recipes.py", line 299, in remote
    return remote.main(args)
  File "/b/.remote_run_cipd/recipe_engine/remote.py", line 44, in main
    args.repository, args.revision, checkout_dir, allow_fetch=True)
  File "/b/.remote_run_cipd/recipe_engine/util.py", line 215, in wrapper
    return f(*args, **kwargs)
  File "/b/.remote_run_cipd/recipe_engine/fetch.py", line 167, in checkout
    git('clone', '-q', repo, checkout_dir)
  File "/b/.remote_run_cipd/recipe_engine/fetch.py", line 142, in __call__
    subcommand, e.message,))
GitError: Git "clone" failed: 
Uncaught exception (GitError): Git "clone" failed: 
 
Blocking: 710128

Comment 2 by pschm...@google.com, Apr 10 2017

Cc: benhenry@chromium.org
The mac_pro_perf_bisect slaves are lowly 11" macbook air's with 120 gb ssd's.
I just clobbered build196-b4 but this is a bandaid.   The other two are close behind.  Maybe it's time to replace these?


[build194-b4] out: /dev/disk1  112Gi  105Gi  6.8Gi    94% 27523091 1775595   94%   /
[build195-b4] out: /dev/disk1  112Gi  109Gi  2.4Gi    98% 28668979 629707   98%   /
[build196-b4] out: /dev/disk1  112Gi   67Gi   44Gi    61% 17689250 11609436   60%   /
Cc: sullivan@chromium.org martiniss@chromium.org
We're going to be putting BattOrs on these, soon. Perhaps we can switch them out at the same time. Are we using other MacBook Airs that aren't explicitly configs with "Air" in their label? Seems odd.
I think something is messed up here. We should have:

mac_pro_perf_bisect: build19{4..6}-b4: MacBook PRO: http://www.everymac.com/systems/apple/macbook_pro/specs/macbook-pro-core-i7-2.5-15-dual-graphics-mid-2015-retina-display-specs.html
It does only have 120GB SSD, but it should NOT be a macbook air.

mac_air_perf_bisect: build(199,200,201)-b4: MacBook AIR: http://www.everymac.com/systems/apple/macbook-air/specs/macbook-air-core-i5-1.6-11-early-2015-specs.html

Did these get mixed up?

Either way, how much disk space is recommended?

Comment 5 by pschm...@google.com, Apr 11 2017

Disk space seems o.k now so it looks like it was a temporary condition.

[build194-b4] out: /dev/disk1  112Gi   89Gi   22Gi    80% 23421449 5877237   80%   /
[build195-b4] out: /dev/disk1  112Gi   80Gi   31Gi    72% 21046287 8252399   72%   /
[build196-b4] out: /dev/disk1  112Gi   81Gi   31Gi    73% 21214619 8084067   72%   /

As for the models, yes they got mixed up:

[build194-b4] out:       Model Identifier: MacBookAir7,1
[build195-b4] out:       Model Identifier: MacBookAir7,1
[build196-b4] out:       Model Identifier: MacBookAir7,1
[build199-b4] out:       Model Identifier: MacBookPro11,5
[build200-b4] out:       Model Identifier: MacBookPro11,5
[build201-b4] out:       Model Identifier: MacBookPro11,5
Well, that certainly makes things complicated...Annie - I'm concerned about all of the regressions with these configs that have caused people to chase their tails. 

Also, if we want to switch them, won't that reset all of the benchmarks that we run on them? 

Comment 7 by d...@chromium.org, Apr 12 2017

Labels: -Infra-Troopers
Recent builds have succeeded. I'm going to remove this from the trooper queue.
Can we check if the main perf waterfall bots are also backwards?

build12{3..7}-b1 should be MacBook AIR
build1{28..23}-b1 build be MacBook PRO
Components: Infra>Labs
Infra>Labs, can you check on #8?
No, those are not backwards.

[build123-b1] out:       Model Identifier: MacBookAir7,1
[build124-b1] out:       Model Identifier: MacBookAir7,1
[build125-b1] out:       Model Identifier: MacBookAir7,1
[build126-b1] out:       Model Identifier: MacBookAir7,1
[build127-b1] out:       Model Identifier: MacBookAir7,1
[build128-b1] out:       Model Identifier: MacBookPro11,5
[build129-b1] out:       Model Identifier: MacBookPro11,5
[build130-b1] out:       Model Identifier: MacBookPro11,5
[build131-b1] out:       Model Identifier: MacBookPro11,5
[build132-b1] out:       Model Identifier: MacBookPro11,5
Components: -Infra>Labs Speed>Benchmarks>Waterfall
Owner: nedngu...@google.com
Ned, can you triage to someone to flip the bisectors? Per comment #5, build194-b4-build196-b4 should be called mac_air_perf_bisect and build199-b4-build201-b4 should be called mac_pro_perf_bisect
Owner: simonhatch@chromium.org
Reassign to Simon since this is related to bisect bots.
Owner: dtu@chromium.org
Dave, can you take a look since it's a holiday in Canada? See #11.
Components: -Speed>Benchmarks>Waterfall Speed>Bisection
Project Member

Comment 15 by bugdroid1@chromium.org, Apr 25 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/tools/build/+/5aeaa309400b74958fe31ae1c6bcb02b5c234ef1

commit 5aeaa309400b74958fe31ae1c6bcb02b5c234ef1
Author: Simon <simonhatch@chromium.org>
Date: Tue Apr 25 14:03:33 2017

Bisect - Macbook air/pro builders got mixed up.

From  crbug.com/710239#c5 

As for the models, yes they got mixed up:

[build194-b4] out:       Model Identifier: MacBookAir7,1
[build195-b4] out:       Model Identifier: MacBookAir7,1
[build196-b4] out:       Model Identifier: MacBookAir7,1
[build199-b4] out:       Model Identifier: MacBookPro11,5
[build200-b4] out:       Model Identifier: MacBookPro11,5
[build201-b4] out:       Model Identifier: MacBookPro11,5


So now:

mac_air_perf_bisect: [194, 195, 196]
mac_pro_perf_bisect: [199, 200, 201]


Bug:  710239 
Change-Id: I2d150741379a93d3d8897dc2881ad107fec338e5
Reviewed-on: https://chromium-review.googlesource.com/485919
Reviewed-by: David Tu <dtu@chromium.org>
Commit-Queue: Simon Hatch <simonhatch@chromium.org>

[modify] https://crrev.com/5aeaa309400b74958fe31ae1c6bcb02b5c234ef1/masters/master.tryserver.chromium.perf/slaves.cfg

Ok switched the builders, just need to restart tryserver.chromium.perf now, anyone know how to do that?
Status: Assigned (was: Untriaged)

Comment 19 by dtu@chromium.org, Aug 22 2017

Status: Fixed (was: Assigned)

Sign in to add a comment