1147835 - SeaMonkey linux* builders are not connected to master due to failing runslave.py's guess_basedir()

Status: RESOLVED FIXED

(SeaMonkey :: Release Engineering, defect)

Description

[Edmund Wong (:ewong)]

With the recent change in runslave.py (http://hg.mozilla.org/build/puppet/rev/0d48d0ca9f54), SeaMonkey's linux* slaves are failing the guess_basedir()
check.

I haven't discovered why it's even getting to that stage yet (still reading
up on runslave.py), but one option is to add seamonkey's linux64 pattern
to the list of slave_match() check.

Comment 1

[Edmund Wong (:ewong)]

Attachment: add SeaMonkey's slaves to check (v1)

this is one option.

Comment 2

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

Perhaps it's time for SM to stop following the moco puppet repository?

Comment 3

[Justin Wood (:Callek)]

(In reply to Dustin J. Mitchell [:dustin] from comment #2)
> Perhaps it's time for SM to stop following the moco puppet repository?

Probably.  However we'd still need to patch this bustage somehow....

Comment 4

[Justin Wood (:Callek)]

Comment on attachment 8583788 [details] [diff] [review]
add SeaMonkey's slaves to check (v1)

Review of attachment 8583788 [details] [diff] [review]:
-----------------------------------------------------------------

I'm going to defer this seamonkey approach to coop, who can tell me if he's ok with us patching this file, or doing anything here.

I have a gut feeling that if we do patch here, we'd want to do sea-* as part of the match, but I'm unsure at present.

Comment 5

[Chris Cooper [:coop] (he/him)]

Comment on attachment 8583788 [details] [diff] [review]
add SeaMonkey's slaves to check (v1)

Review of attachment 8583788 [details] [diff] [review]:
-----------------------------------------------------------------

(In reply to Justin Wood (:Callek) from comment #4) 
> I'm going to defer this seamonkey approach to coop, who can tell me if he's
> ok with us patching this file, or doing anything here.
> 
> I have a gut feeling that if we do patch here, we'd want to do sea-* as part
> of the match, but I'm unsure at present.

I agree with Dustin: SeaMonkey should be maintaining a fork of this code and updating it on their own schedule to avoid exactly these types of issues.

Comment 6

[Edmund Wong (:ewong)]

Since this no longer touches relops' puppet, moving to SeaMonkey::RelEng.

Comment 7

[Edmund Wong (:ewong)]

Attachment: change repo url in seamonkey-config.pp (v1)

Comment 8

[Edmund Wong (:ewong)]

Comment on attachment 8590633 [details] [diff] [review]
change repo url in seamonkey-config.pp (v1)

Pushed to SeaMonkey/puppet:
http://hg.mozilla.org/SeaMonkey/puppet/rev/8059d077b351

Comment 9

[Edmund Wong (:ewong)]

Pushed to SeaMonkey/puppet the backout of cset 0d48d0ca9f54:
http://hg.mozilla.org/SeaMonkey/puppet/rev/7e6b67b6a53d

+ modified the .hg/hgrc in seapuppet's /etc/puppet/production to
reflect the new repo http://hg.mozilla.org/SeaMonkey/puppet.

Now,  the linux slaves are starting to connect with the master again.

Comment 10

[Edmund Wong (:ewong)]

Comment on attachment 8583788 [details] [diff] [review]
add SeaMonkey's slaves to check (v1)

No longer need this patch.

Comment 11

[Edmund Wong (:ewong)]

The linux* slaves are back (minus -5 and -7 due to them being AWOLish).

Closing this bug.