877342 - Kill mozilla-esr17 (2 versions after FF24 ships)

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[Armen [:armenzg]]

I believe FF24 will ship on the 10th of September and FF26 on 3rd of December.

At that point we will stop running rev3 jobs from esr17.

Comment 1

[[e-mail] (:bhearsum)]

esr17 is long dead

Comment 2

[Phil Ringnalda (:philor)]

You must be thinking of esr10; esr17 isn't dead until next December.

Comment 3

[Armen [:armenzg]]

Merge day is on Dec. 9th as per https://wiki.mozilla.org/RapidRelease/Calendar

I will help drive this.

Comment 4

[Armen [:armenzg]]

Attachment: [configs] remove esr17

Comment 5

[Armen [:armenzg]]

Attachment: [buildbotcustom] remove esr17

Comment 6

[Armen [:armenzg]]

Attachment: [tools] remove esr17

Comment 7

[Armen [:armenzg]]

Attachment: list of Rev3 Windows builders after these patches

Comment 8

[Armen [:armenzg]]

Attachment: these are the Windows test builders that differ prior to the current patches

Comment 10

[Armen [:armenzg]]

Attachment: [configs] remove talos-r3-xp

Once the changes are live:
* remove all xp machines from slavealloc
* request IT to decommission all talos-r3-xp-### machines

Comment 11

[[e-mail] (:bhearsum)]

Comment on attachment 8343968 [details] [diff] [review]
[configs] remove esr17

Review of attachment 8343968 [details] [diff] [review]:
-----------------------------------------------------------------

I tried to run dump_masters.sh on this but I couldn't get it to produce terribly useful output. I'm going to trust that this doesn't break anything, but the changes to the default set of tests are a little scary.

::: mozilla-tests/config.py
@@ -86,5 @@
>              'linux64': {},
>          },
>          'lock_platforms': True,
>      },
> -    'mozilla-b2g18_v1_0_1': {

Did you mean to remove this branch too? If so, it probably deserves its own bug. In the future I'd lean towards changing the platform defaults separately too -- we don't want to be forced to re-enable esr17 builds/builders if something goes wrong with that code. I'm not going to make you re-do these patches for that though.

::: test-masters.sh
@@ +15,5 @@
>  # I have had problems where a whole bunch of parallel HTTP requests caused
>  # errors (?), so fetch it once here and pass it in.
>  MASTERS_JSON=$(mktemp $WORK/tmp.masters.XXXXXXXXXX)
> +#wget -q -O$MASTERS_JSON "$MASTERS_JSON_URL" || exit 1
> +cp $MASTERS_JSON_URL $MASTERS_JSON || exit 1

I don't think you want to land this part.

Comment 12

[[e-mail] (:bhearsum)]

Additional things to kill:
- update verify configs
- patcher configs
- l10n changesets
- l10n mozconfigs (in buildbot-configs)

Comment 13

[Armen [:armenzg]]

wrt mozilla-b2g18_v1_0_1, we don't run anything there. You can see that it is not on tbpl.
wrt to the platform default changes, esr17 is the last repo that uses the xp machines, hence, I deleted it together.

I will look for the code mentioned on comment 12.

Comment 14

[Armen [:armenzg]]

Comment on attachment 8343970 [details] [diff] [review]
[tools] remove esr17

https://hg.mozilla.org/build/tools/rev/482e03f6ca02

Comment 15

[Armen [:armenzg]]

Comment on attachment 8343968 [details] [diff] [review]
[configs] remove esr17

I can see that I attached the patch that had areas commented out. I landed the right thing (some uncommenting and loop fixing).

I run passed test-masters.sh just in case.

https://hg.mozilla.org/build/buildbot-configs/rev/0b2a63a1c371

Comment 16

[Armen [:armenzg]]

Comment on attachment 8343969 [details] [diff] [review]
[buildbotcustom] remove esr17

https://hg.mozilla.org/build/buildbotcustom/rev/ad6c96f6e4cc

Comment 17

[Aki Sasaki (not active)]

Landed http://hg.mozilla.org/build/buildbot-configs/rev/f33e13a9f9c7 to turn fedora talos back off, since that was closing all trees.

Comment 18

[hwine]

in production

Comment 19

[Armen [:armenzg]]

Thank you!
Sorry for the removal of that block. The comment mislead me.

Comment 20

[Armen [:armenzg]]

Attachment: [configs] remove esr17 files

Let's clean this as well.

Comment 21

[Armen [:armenzg]]

Attachment: [tools] remove esr17 files

Comment 22

[Nick Thomas [:nthomas] (UTC+12)]

(In reply to Armen Zambrano [:armenzg] (Release Engineering) (EDT/UTC-4) from comment #10)
> Created attachment 8344012 [details] [diff] [review]
> [configs] remove talos-r3-xp
> 
> Once the changes are live:
> * remove all xp machines from slavealloc
> * request IT to decommission all talos-r3-xp-### machines

These machines are still trying to log into masters, but the masters no longer accepts them, so we get a lot of 
  	Failure: twisted.cred.error.UnauthorizedLogin: 
in the master exception emails. To the tune of several MB per master per hour.

Please disable them in slavealloc, and reboot the machines, asap.

Comment 23

[Armen [:armenzg]]

FTR coop has dealt with disabling them and rebooting them.

Comment 24

[[e-mail] (:bhearsum)]

Comment on attachment 8345369 [details] [diff] [review]
[configs] remove esr17 files

Review of attachment 8345369 [details] [diff] [review]:
-----------------------------------------------------------------

r=me if you don't touch calendar configs. Those aren't ours.

Comment 25

[Armen [:armenzg]]

Comment on attachment 8345369 [details] [diff] [review]
[configs] remove esr17 files

I left calendar files untouched:
https://hg.mozilla.org/build/buildbot-configs/rev/42503b049a4e

Comment 26

[Armen [:armenzg]]

Comment on attachment 8345376 [details] [diff] [review]
[tools] remove esr17 files

https://hg.mozilla.org/build/tools/rev/7d84050de730

Comment 27

[Armen [:armenzg]]

* removed slaves from slavealloc
* disabled them from running on kittenherder's crontab

mysql> select count(*) from slaves where name like 'talos-r3-xp-%';
+----------+
| count(*) |
+----------+
|      139 |
+----------+
1 row in set (0.23 sec)

mysql> delete from slaves where name like 'talos-r3-xp-%';
Query OK, 139 rows affected (0.20 sec)

mysql> select count(*) from slaves where name like 'talos-r3-xp-%';
+----------+
| count(*) |
+----------+
|        0 |
+----------+
1 row in set (0.19 sec)

Comment 28

[[e-mail] (:bhearsum)]

Something here is in production