781286 - duplicate entries in staging buildbot configs for r5 slaves

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: CIDuty, task, P3)

Description

[[e-mail] (:bhearsum)]

I see two entries for 090 and some other slaves on my staging master.

Comment 1

[Rail Aliiev [:rail]]

Hmm, 
http://hg.mozilla.org/build/buildbot-configs/rev/5aba8ed85758 added new lion build slaves, but some of them are still marked as "dev/pp" in slavealloc...

Comment 2

[Rail Aliiev [:rail]]

Attachment: rm staging lion slaves

Comment 3

[[e-mail] (:bhearsum)]

Can we make sure that bulidbot + slavealloc agree on where these slaves should be?

Comment 4

[Rail Aliiev [:rail]]

Comment on attachment 650318 [details] [diff] [review]
rm staging lion slaves

http://hg.mozilla.org/build/buildbot-configs/rev/216895ff8a49

New tests will be ready soon

Comment 5

[Rail Aliiev [:rail]]

Attachment: tests

* Use STAGING_SLAVES variable to simplify the tests
* Drop test_preprod_vs_stag: we don't use a separate list for preproduction anymore
* Drop test_prod_is_subset_of_stag: the current config extends staging slave list
* Add test_stag_not_in_prod: check that staging-only slaves and production slaves have nothing in common.

Comment 6

[Rail Aliiev [:rail]]

Comment on attachment 650864 [details] [diff] [review]
tests

http://hg.mozilla.org/build/buildbot-configs/rev/b6021cb9a84c

Comment 7

[Rail Aliiev [:rail]]

Comment on attachment 650864 [details] [diff] [review]
tests

http://hg.mozilla.org/build/buildbot-configs/rev/719e8f4486f8

Comment 8

[Rail Aliiev [:rail]]

It breaks mozilla-tests/test. Needs a workaround.

Comment 9

[Rail Aliiev [:rail]]

Attachment: tests v2

I refactored mozilla-tests/{stagin,preproduction}_config.py to not duplicate slave entries and meet new tests.

Comment 10

[Rail Aliiev [:rail]]

Comment on attachment 654232 [details] [diff] [review]
tests v2

http://hg.mozilla.org/build/buildbot-configs/rev/6681c042682f

Comment 11

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

This made it into production today.

Comment 12

[Aki Sasaki (not active)]

I think this broke new staging master creation, at the very least.

Comment 13

[Aki Sasaki (not active)]

Attachment: staging_config.SLAVES pprint, before (rev 14a926adf2ea)

This is the output from 14a926adf2ea when I put a

import pprint
pprint.pprint(SLAVES)

in staging_config.py.  Notice it's a dict, which is important for http://hg.mozilla.org/build/buildbot-configs/file/a6480010748d/mozilla-tests/universal_master_sqlite.cfg#l75

Comment 14

[Aki Sasaki (not active)]

Attachment: staging_config.SLAVES pprint, after (rev 6681c042682f)

This is after; SLAVES is a list of tuples, breaking slaves.keys().

Comment 15

[Aki Sasaki (not active)]

Attachment: fix mozilla-tests staging slaves

Comment 16

[Rail Aliiev [:rail]]

Comment on attachment 657492 [details] [diff] [review]
fix mozilla-tests staging slaves

Ahhhh, right, I forgot about dict() :/

Comment 17

[Aki Sasaki (not active)]

Comment on attachment 657492 [details] [diff] [review]
fix mozilla-tests staging slaves

http://hg.mozilla.org/build/buildbot-configs/rev/e708e61e163b

Comment 18

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

Comment on attachment 657492 [details] [diff] [review]
fix mozilla-tests staging slaves

This is in production now.