1370877 - [generic-worker] log problems uploading artifacts to the task log

Status: RESOLVED FIXED

(Taskcluster :: Workers, enhancement)

Description

[Dustin J. Mitchell [:dustin] (he/him)]

https://tools.taskcluster.net/task-inspector/#dTvuIh14RBuz2v-HYk9zWw/

ends with

[taskcluster 2017-06-06T21:21:00.371Z]   Exit Code: 0
[taskcluster 2017-06-06T21:21:00.371Z]   User Time: 0s
[taskcluster 2017-06-06T21:21:00.371Z] Kernel Time: 0s
[taskcluster 2017-06-06T21:21:00.371Z]   Wall Time: 2m23.4749969s
[taskcluster 2017-06-06T21:21:00.372Z] Peak Memory: 1720320
[taskcluster 2017-06-06T21:21:00.372Z]      Result: SUCCEEDED
[taskcluster 2017-06-06T21:21:00.372Z] === Task Finished ===
[taskcluster 2017-06-06T21:21:00.372Z] Task Duration: 6m22.6110922s

yet the task status is "failure".  The guess is, something went awry with uploading of artifacts, but there's no logging of what that might have been.

Comment 1

[Pete Moore [:pmoore][:pete]]

This is fixed in later generic-worker version, which should be rolled out this week everywhere.

Comment 2

[Pete Moore [:pmoore][:pete]]

It looks like this was included in generic-worker 8.5.0. These worker types need upgrading (which is already underway in bug 1373722):

  1) aws-provisioner-v1/gecko-t-win7-32       (generic-worker 8.2.0)
  2) aws-provisioner-v1/gecko-t-win7-32-gpu   (generic-worker 8.2.0)
  3) aws-provisioner-v1/gecko-t-win10-64      (generic-worker 8.3.0)


The remaining mainstream generic-worker worker types are already sufficiently up-to-date (i.e. >= 8.5.0):

  1) releng-hardware/gecko-t-osx-1010         (generic-worker 8.5.0)
  2) aws-provisioner-v1/gecko-1-b-win2012     (generic-worker 10.0.5)
  3) aws-provisioner-v1/gecko-t-win10-64-gpu  (generic-worker 10.0.5)
  4) aws-provisioner-v1/nss-win2012r2         (generic-worker 10.0.5)
  5) aws-provisioner-v1/win2012r2             (generic-worker 10.0.5)

Comment 3

[Greg Arndt [:garndt]]

I think we've long since upgraded these workers and this is fixed?

Comment 4

[Pete Moore [:pmoore][:pete]]

(In reply to Greg Arndt [:garndt] from comment #3)
> I think we've long since upgraded these workers and this is fixed?

Sadly not! :(

But soon!

Comment 5

[Pete Moore [:pmoore][:pete]]

Fixed with the rollout in bug 1399401.