663615 - builds-4hr.js.gz is missing slaves

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[Markus Stange [:mstange]]

Attachment: offending builds-4hr.js

Today the import script used in the future buildbot-based TBPL couldn't get slave names for some builds because their slave_id wasn't present in the slaves list in the JSON at http://build.mozilla.org/builds/builds-4hr.js.gz .
For example, build 5054074 (Rev3 WINNT 5.1 try opt test mochitests-3/5) has slave_id 1289 which doesn't exist.
(It *does* exist in http://build.mozilla.org/builds/builds-2011-06-11.js.gz where it maps to "talos-r3-xp-054".)

TBPL needs the slave name when communicating with the WOO orange factor database. Slave names are shown in the Test Machine column for example on http://brasstacks.mozilla.com/orangefactor/?display=Bug&endday=2011-06-11&startday=2011-04-12&bugid=622059

Comment 1

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

This may be unrelated, but just to check: if new slaves *do* appear in builds-4hr.js.gz, will those correctly appear in the WOO db?  We already have a bit of a hard time adding slaves because they must be added to the graphs database, and this impedes our ability to adjust the slave pool to meet demand.

Comment 2

[Markus Stange [:mstange]]

(In reply to comment #1)
> This may be unrelated, but just to check: if new slaves *do* appear in
> builds-4hr.js.gz, will those correctly appear in the WOO db?

I don't know, but jgriffin probably does.

I'd expect it to work, though, because TBPL sends slave _names_ to the WOO db, not slave IDs.

Comment 3

[Jonathan Griffin (:jgriffin)]

(In reply to comment #2)
> (In reply to comment #1)
> > This may be unrelated, but just to check: if new slaves *do* appear in
> > builds-4hr.js.gz, will those correctly appear in the WOO db?
> 
> I don't know, but jgriffin probably does.
> 
> I'd expect it to work, though, because TBPL sends slave _names_ to the WOO
> db, not slave IDs.

This shouldn't cause any problems with WOO.

Comment 4

[Chris AtLee [:catlee]]

Is this happening consistently, or only on occasion?

The slave ids are static, so you should be able to remember them between runs.

Comment 5

[Chris AtLee [:catlee]]

I made a small change to our reporting that should fix this. Let me know if you hit it again.

Comment 6

[Markus Stange [:mstange]]

(In reply to comment #4)
> Is this happening consistently, or only on occasion?

Only on occasion. I haven't seen it happen again after the day I filed this bug.