New issue
Advanced search Search tips

Issue 708848 link

Starred by 3 users

Issue metadata

Status: Verified
Owner:
Closed: Apr 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: Bug



Sign in to add a comment

Sheriffing reminder emailer on master1 isn't working for perf.rotations.json

Project Member Reported by benhenry@chromium.org, Apr 6 2017

Issue description

Doc:
https://chrome-internal.googlesource.com/infra/infra_internal/+/master/infra_internal/services/sheriff/

Relevant quote:
"Note, shift reminder emails will automatically be sent (via the chrome-bot crontab on master1.golo.chromium.org), unless you disable them in your rotation config, as described below."

rotation:
https://chrome-internal.googlesource.com/infra/infra_internal/+/master/infra_internal/services/sheriff/rotations/perf.rotation.json

Looks like master1 isn't doing what it's expected to do - maybe this has to do with the recent split. 
 
Owner: mmoss@chromium.org
Status: Assigned (was: Untriaged)
This looks like the cron entry:

# Send sheriff reminder emails every day.
MAILTO="mmoss@google.com"
0 3 * * Mon-Fri /home/chrome-bot/buildbot/build_internal/scripts/tools/sheriff/reminder_emailer.py --oauth_file /home/chrome-bot/buildbot/build_internal/scripts/tools/sheriff/.sheriff_oauth --message_from_address chrome-team-noreply@chromium.org --config chrome --config bling_europe --config bling_us --config chrome --config chromeos --config chromeos_lab --config chromeos.other --config chromeos_test --config clank --config clank_merge --config gpu --config nacl --config network --config troopers --config valgrind --config webkit --config perf --loglevel error

Errors are mailed to mmoss, unfortunately I think that means there are no error logs on the master.

Comment 2 by mmoss@chromium.org, Apr 10 2017

Are you sure that's the right cron entry? I thought we got rid of all the separate "--config" args, so we didn't have to change cron every time we added/removed a rotation. Also, I haven't gotten any cron failure emails.
Cc: mmoss@chromium.org
Owner: ----
Status: Untriaged (was: Assigned)

Comment 4 by mmoss@chromium.org, Apr 10 2017

Cc: hinoka@chromium.org
OK, that cron is definitely old. hinoka@, where did you come up with that? If that's still live somewhere, it should really get cleaned up.

Here's the right one:

https://chrome-internal.googlesource.com/infra/puppet.git/+/master/puppetm/etc/puppet/modules/chrome_infra/files/setup/cron/master1.golo.chromium.org/chrome-infra-cron#82

and looks like emails should go to chrome-golo-alerts@google.com
It could be caused by the recent master split. I remember being unsure about whether or not we needed to copy those over.

Comment 6 by hinoka@chromium.org, Apr 10 2017

crontab -e was empty, but i saw a cron file in the homedir and thought that it might be in use.  If it is infact not the right cron file, then it should probably be removed to avoid further confusing.

Comment 7 by hinoka@chromium.org, Apr 10 2017

confusion*

Comment 8 by d...@chromium.org, Apr 12 2017

Owner: hinoka@chromium.org
Status: Assigned (was: Untriaged)
Assigning to hinoka@, since he's been looking into this. Please re-add to trooper queue with a status update if you are no longer working on this, or close if it's fixed!

Comment 9 by mmenke@chromium.org, Apr 19 2017

Cc: mmenke@chromium.org
Labels: -Pri-2 Pri-1
We're having the same issue with network.rotation.json.  It looks like it may have coincided with extending the rotation.  No one's been sheriffing for the past 9 days as a result.  This seems like a P-1 to me.

Comment 10 by jpar...@google.com, Apr 19 2017

Same issue for monorail rotation, causing rotation to be missed

Hinoka, any update?
Owner: benhenry@chromium.org
Assigning to benhenry@, the listed owner for sheriff rotations.  Can you take a look?
https://chrome-internal.googlesource.com/infra/infra_internal/+/master/doc/services.md#sheriff-trooper-rotation

I don't know anything about a master split.
Owner: martiniss@chromium.org
Nope, sorry. How about Stephen.
I'm not an owner of this. I'll look at this though.
Status: Started (was: Assigned)
I think it's a bad config. CL out: https://chrome-internal-review.googlesource.com/c/357943/
Project Member

Comment 16 by bugdroid1@chromium.org, Apr 19 2017

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal/+/7cf1479a586e27aca8157faf3d28d5ddaef720b0

commit 7cf1479a586e27aca8157faf3d28d5ddaef720b0
Author: Stephen Martinis <martiniss@google.com>
Date: Wed Apr 19 22:09:57 2017

The cron should run again tonight (3 AM PST), and it should pass this time. The thread linked in #14 will get another email if it fails again.
Can one bad rotation config file make other rotations stop sending emails?  I assume that file's only used when updating that particular rotation, and not parsed when updating other rotations.
Owner: ----
Status: Fixed (was: Started)
I'm pretty sure it'll stop all rotations from sending emails. The logic for sending emails is here: https://chrome-internal.googlesource.com/infra/infra_internal.git/+/master/infra_internal/services/sheriff/reminder_emailer.py

It loops through all known configs. If it hits an error, I think it fails.

mmoss@ would know more though. He wrote that code.
Thanks!  Makes sense, I just assumed all of that was configured on rotation update, and integrated into gcal's reminder feature somehow.
Project Member

Comment 21 by bugdroid1@chromium.org, Apr 20 2017

The following revision refers to this bug:
  https://chrome-internal.googlesource.com/infra/infra_internal/+/3dcf8425f5eb0156fb92cf19f3bdc40f91de2fb6

commit 3dcf8425f5eb0156fb92cf19f3bdc40f91de2fb6
Author: John Budorick <jbudorick@google.com>
Date: Thu Apr 20 00:08:13 2017

Owner: martiniss@chromium.org
Status: Verified (was: Fixed)
Components: -Infra Infra>Sheriffing>Rotations

Sign in to add a comment