Closed Bug 1131787 Opened 9 years ago Closed 7 years ago

production-ify allthethings.json

Categories

(Release Engineering :: General, defect)

x86_64
Linux
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: catlee, Unassigned)

References

Details

Attachments

(2 files, 1 obsolete file)

We need a real home for the tools that generate allthethings.json to live if we're going to be relying on it in production. It also needs proper error reporting, nagios checks, etc.
I created a wiki page (until this is done) with minimal information I know about:
https://wiki.mozilla.org/ReleaseEngineering/How_To/allthethings.json
Step 1: move code into buildbot-configs repo
Attached patch bug1131787.patch (obsolete) — Splinter Review
Attachment #8622641 - Flags: review?(catlee)
Assignee: nobody → sdeckelmann
First patch moves files into buildbotcustom repo.
Comment on attachment 8622641 [details] [diff] [review]
bug1131787.patch

Review of attachment 8622641 [details] [diff] [review]:
-----------------------------------------------------------------

lgtm.

does make_allthethings.sh need to be part of this, or should that go in the deployment code? it has fairly specific paths in it.
Attachment #8622641 - Flags: review?(catlee) → review+
Moved make_allthethings.sh into puppet config and cron under 'buildduty' and a few minor cosmetic updates to scripts. (update to other patch coming in a moment)
Attachment #8622688 - Flags: review?(catlee)
r+ from catlee - moved one file into puppet.
Attachment #8622641 - Attachment is obsolete: true
Comment on attachment 8622688 [details] [diff] [review]
bug1131787-puppet.patch

Review of attachment 8622688 [details] [diff] [review]:
-----------------------------------------------------------------

Adding in the comments from irc. We want some failsafes around things like simultaneous run from cron (use lockfile), cleaning up lockfiles/tempp file when interrupted (use trap), making sure we put timeouts in the hg commands so they don't hang forever, and sending the output to syslog instead of cronmail (use logger -t).

I mentioned that looking at the crontask class might be a good fit (most of the aws cron jobs use this, see most of the lines above that one in the cron.pp file) but maybe not everything lines up there.
Attachment #8622688 - Flags: review?(catlee) → review+
I was asked to use 'hgtool.py' for hg timeouts. working on it.
Hi Selena, are you still working on this?
Flags: needinfo?(sdeckelmann)
(In reply to Armen Zambrano Gasparnian [:armenzg] from comment #10)
> Hi Selena, are you still working on this?

Ah - I was just thinking about this yesterday. I can poke at it tomorrow and let you know whether or not I can get it into production.
Oops, pushing to next week. Sorry!
Flags: needinfo?(sdeckelmann)
Assignee: sdeckelmann → nobody
We have nagios checks on builddata.pub.build.mozilla.org on /buildjson/builds-4hr.js.gz and friends, so that's a natural place to add a new check for /reports/allthethings.json and
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → WORKSFORME
Component: General Automation → General
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: