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

Issue 901236 link

Starred by 4 users

Issue metadata

Status: Fixed
Owner:
Last visit 26 days ago
Closed: Nov 13
Cc:
EstimatedDays: ----
NextAction: ----
OS: Linux , Windows , Mac
Pri: 1
Type: Bug-Regression



Sign in to add a comment

Runtime error in - per revision bisect script for certain revisions.

Project Member Reported by nyerramilli@chromium.org, Nov 2

Issue description

We are getting "RuntimeError: We don't have enough builds to bisect. revlist: []" for certain revision ranges in All OS.

Few examples: (We tried in different systems, and also increased the build ranges but got the same error)

Good build: 72.0.3596.0 (Revision: 603700)
Bad build : 72.0.3597.0 (Revision: 604362)

Good build: 72.0.3588.0 (Revision: 601409)
Bad build : 72.0.3589.0 (Revision: 601810)

Good build: 72.0.3579.0 (Revision: 599392)
Bad build : 72.0.3580.0 (Revision: 599523)

Good Build : 71.0.3568.0 (Revision : 595545)
Bad Build : 71.0.3569.0 (Revision : 596015)

Screenshot attached for reference.

Prasad, could you please check and help.
 
bisect script err -runtime.jpg
199 KB View Download
This might be related to https://crbug.com/893686, recent migration of perf-builder to LUCI broke archiving builds, this got fixed now, but unfortunately builds are missing for this period.
Also I noticed that there was some changes to folder names in GS, I uploaded a new version bisect_builds.py. Please download it from here https://sites.google.com/a/google.com/chrome-te/home/tools/bisect_builds?pli=1 and re-run the bisect.
Thanks prasadv@ for fixing this. This seems to be working fine now with the new version bisect_builds.py on Mac in the range: Good build: 72.0.3588.0 (Revision: 601409)
Bad build : 72.0.3589.0 (Revision: 601810). Earlier, I was facing the error in the same regression range on Mac.  

We will test this on other OS as well and update the behavior.
Cc: viswa.karala@chromium.org
Labels: Needs-Feedback
Tested the per-revision bisect on Windows, Mac and Linux using new version of bisect_builds.py with M-71 and M-72. For these two milestones with new version of bisect_builds.py it's working fine and builds are getting invoked.But, with the same new version of bisect_builds.py for M-70 milestone we are facing error on running per-revision bisect.

Example:
Good: 70.0.3503.0
Bad: 70.0.3504.0
Error: RuntimeError: We don't have enough builds to bisect. revlist: []

Good: 70.0.3502.0
Bad: 70.0.3503.0
Error: RuntimeError: We don't have enough builds to bisect. revlist: []

Good: 70.0.3537.0
Bad: 70.0.3538.0
Error: [Errno 2] No such file or directory

prasadv@ Could you please confirm that we need to raise a separate issue for the same and also is there any issue in running the earlier ranges with the new version of bisect_builds.py

Thanks!
Thanks for the fix in comment 2!  I was hitting this on Windows and doing a "git pull" in my bisect_builds directory (to pick up the fix) resolved it for me.  (I'm bisecting within M72, so I'm not sure about the M70 issue in comment 4.)
Labels: -Needs-Feedback
We are facing error (comment# 4) for M70 versions of chrome, works fine in M71 and M72. 

Prasad, would you mind taking a look.
Project Member

Comment 7 by bugdroid1@chromium.org, Nov 13

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/chrome/test/python_tests/+/624e1bbf134983c382f617b00d051f64cb53a49a

commit 624e1bbf134983c382f617b00d051f64cb53a49a
Author: Prasad Vuppalapu <prasadv@google.com>
Date: Tue Nov 13 22:58:27 2018

Status: Fixed (was: Assigned)
Please checkout the latest version of bisect_build.py to verify this.

Issue 916325 has been merged into this issue.

Sign in to add a comment