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

Issue 922993 link

Starred by 2 users

Issue metadata

Status: Fixed
Owner:
Closed: Today
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 1
Type: Bug



Sign in to add a comment

moblab-vm-generic-paladin: host did not return from reboot

Project Member Reported by semenzato@chromium.org, Jan 17 (5 days ago)

Issue description

This broke a CQ run.  

This build failed because (I think) some provision step in moblab-vm autotests did not return from reboot.

The only other error was on gale-paladin which was set as experimental in the tree status.

https://ci.chromium.org/p/chromeos/builders/luci.chromeos.general/Prod/b8924093670230016256

I found a CIDB failure in this log, but I don't know if it's related.

https://logs.chromium.org/logs/chromeos/buildbucket/cr-buildbucket.appspot.com/8924093670230016256/+/steps/CommitQueueCompletion/0/stdout

04:43:00: INFO: Build config moblab-generic-vm-paladin completed: CIDB status: fail. Buildbucket status STARTED result None.

Then I followed the moblab-vm-generic-paladin link and found this:

https://ci.chromium.org/p/chromeos/builders/luci.chromeos.general/CQ/b8924092351205464368

and from there I found this moblab test log:

https://logs.chromium.org/logs/chromeos/buildbucket/cr-buildbucket.appspot.com/8924092351205464368/+/steps/MoblabVMTest/0/stdout

which first shows an error getting the HWIDLabel:

04:16:34 INFO | autoserv| Retrying in 13.670080 seconds...
04:17:00 INFO | autoserv| Setting image_storage_server to /mnt/moblab/static/prefetched/
04:17:09 INFO | autoserv| [stderr] ERROR:root:error getting label HWIDLabel.

and shortly later:

04:34:11 INFO | autoserv| [stderr] 01-17-2019 [04:34:11] Suite job is finished.
04:34:11 INFO | autoserv| [stderr] 01-17-2019 [04:34:11] Start collecting test results and dump them to json.
04:34:12 INFO | autoserv| [stderr] Suite job   [ PASSED ]
04:34:12 INFO | autoserv| [stderr] provision   [ FAILED ]
04:34:12 INFO | autoserv| [stderr] provision     ABORT: Host did not return from reboot

I am guessing that's the problem.
 

Comment 1 by jclinton@google.com, Jan 17 (5 days ago)

Labels: OS-Chrome
Owner: haddowk@chromium.org
Status: Assigned (was: Untriaged)

Comment 2 by wtlee@chromium.org, Yesterday (45 hours ago)

Labels: -Pri-2 Pri-0
Keep seeing these error messages on moblab-vm-generic-paladin builder which keeps blocking CQ. Marking moblab-vm-generic-paladin to experimental seems not work. Raised the priority to 0. 

Comment 3 by wtlee@chromium.org, Yesterday (44 hours ago)

Cc: wtlee@chromium.org

Comment 4 by semenzato@chromium.org, Today (14 hours ago)

Isn't it the case that moblab-vm-generic-paladin could not break the CQ because it was set as EXPERIMENTAL in http://chromiumos-status.appspot.com?

Comment 5 by semenzato@chromium.org, Today (14 hours ago)

Status: WontFix (was: Assigned)
Please correct me if I am wrong.  Also, I saw a CL which makes it experimental permanently.

Comment 6 by semenzato@chromium.org, Today (13 hours ago)

Cc: semenzato@chromium.org
Status: Fixed (was: WontFix)
The CL is https://crrev.com/c/1424718

Also, I just noticed the comments on crosoncall which state that EXPERIMENTAL= doesn't always work.  (Do we have a bug for that?)

Comment 7 by lannm@google.com, Today (13 hours ago)

> Also, I just noticed the comments on crosoncall which state that EXPERIMENTAL= doesn't always work.

Can you clarify your expectations here? The only place that EXPERIMENTAL= has an effect is in the status at chromiumos-status.appspot.com

Comment 8 by lannm@google.com, Today (13 hours ago)

Never mind, I misunderstood the comment. I have read the crosoncall thread.

I'm not aware of the EXPERIMENTAL= mechanism being flaky per se except that the code that makes it work is quite strict about the format of the status message. If you believe there is an instance of it not working then please open a bug.

Comment 9 by wtlee@google.com, Today (12 hours ago)

Labels: -Pri-0 Pri-1
Ohh, sorry for being unclear. Originally, I thought the EXPERIMENTAL= not work since the builder does not show experimental mark "*" next to their name in legoland. But finally I found out that only the builder being set to experimental permanently will have "*" mark. And the EXPERIMENTAL= actually works on the build result.

Currently, it is not moblab-vm-generic-paladin block the CQ, but it keeps failing in latest builds. Please also help to suggest if you know how we can solve the error. 

The logs are: https://luci-logdog.appspot.com/logs/chromeos/buildbucket/cr-buildbucket.appspot.com/8923642069053448480/+/steps/MoblabVMTest/0/stdout

If you think it would be better to create a new bug for this issue, please let me know. Thanks.

Comment 10 by wtlee@google.com, Today (12 hours ago)

And I think we can remove moblab-vm-generic-paladin from being experimental on tree status and revert the CL which makes it experimental once we make those failing tests passed. 

Comment 11 by semenzato@chromium.org, Today (12 hours ago)

#9 thank you for clarifying.  Please look for existing moblab-vm-generic bugs that may be covering the issue you're seeing, and if you don't find one, or are not sure, open another bug.

Sign in to add a comment