929288 - Reconfigure alder builds

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[Joshua Cranmer [:jcranmer]]

I've reserved alder for the purpose of driving the final changes in bug 648979. As a result, it needs to be configured with what should hopefully be the final release configurations for comm-central.

I don't know the magic keywords for release engineering, so feel free to ask me if you have questions about what I mean, but I think the desired config is as follows:
Build on Linux32, Linux64, OS X, Win32 in opt and debug variations, using relevant mozconfigs from mail/ instead of browser/
Tests on OS X 10.6, 10.7, Windows XP and Windows 7. Only the xpcshell tests and the mozmill tests are needed

Ideally, all of the configurations should be as close to those used for m-c based trees with the exception of the different mozconfigs.

Comment 1

[Justin Wood (:Callek)]

Attachment: [configs] support alder for TB

This patch does a few things, it adds project branch support for TB *builds*, it configs alder to be enabled for TB, and it configures alder to have an empty 'mozilla dir' because its a branch with m-c and c-c merged together.

It does *not* yet configure tests, though this setup should be enough to unblock joshua today, and the tests will be required to finish this work.

A note that I'll have to do when deploying is making sure to symlink the new thunderbird_project* before a reconfig though

Comment 2

[Justin Wood (:Callek)]

Comment on attachment 8337755 [details] [diff] [review]
[configs] support alder for TB

https://hg.mozilla.org/build/buildbot-configs/rev/830cc1276096

Comment 3

[Justin Wood (:Callek)]

something[s] here made it to production

Comment 4

[Joshua Cranmer [:jcranmer]]

The opt Linux and Linux64 builds both appear to be purple--trying to post to the graph server is failing with "No machine_name called 'TB_Linux_x86-64_alder' can be found.

Linux and Linux64 debug are both green (yay), and OS X and Windows are red because, well, that's why I requested the tree :-)

Comment 5

[Justin Wood (:Callek)]

I *think* this is fixed graphserver wise, retriggering purple jobs to test:

mysql> insert into machines values (NULL,6,0,NULL,"TB_Linux_alder",1,unix_timestamp());
Query OK, 1 row affected (0.02 sec)

mysql> insert into machines values (NULL,6,0,NULL,"TB_Linux_alder_leak_test",1,unix_times
tamp());
Query OK, 1 row affected (0.01 sec)

mysql> insert into machines values (NULL,18,0,NULL,"TB_Linux_x86-64_alder",1,unix_timesta
mp());
Query OK, 1 row affected (0.00 sec)

mysql> insert into machines values (NULL,18,0,NULL,"TB_Linux_x86-64_alder_leak_test",1,un
ix_timestamp());
Query OK, 1 row affected (0.01 sec)

mysql> insert into machines values (NULL,7,0,NULL,"TB_OS_X_10.7_32-bit_alder_leak_test",1
,unix_timestamp());
Query OK, 1 row affected (0.01 sec)

mysql> insert into machines values (NULL,17,0,NULL,"TB_OS_X_10.7_alder",1,unix_timestamp(
));
Query OK, 1 row affected (0.01 sec)

mysql> insert into machines values (NULL,8,0,NULL,"TB_WINNT_5.2_alder",1,unix_timestamp()
);
Query OK, 1 row affected (0.01 sec)

mysql> insert into machines values (NULL,8,0,NULL,"TB_WINNT_5.2_comm-aurora_leak_test",1,
unix_timestamp());
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect...
Connection id:    4695640
Current database: graphs_stage2_prod

ERROR 1062 (23000): Duplicate entry 'TB_WINNT_5.2_comm-aurora_leak_test' for key 'name'
mysql> insert into machines values (NULL,8,0,NULL,"TB_WINNT_5.2_alder_leak_test",1,unix_t
imestamp());
Query OK, 1 row affected (0.64 sec)

mysql> insert into machines values (NULL,19,0,NULL,"TB_WINNT_6.1_alder",1,unix_timestamp(
));
Query OK, 1 row affected (0.62 sec)

mysql> insert into machines values (NULL,19,0,NULL,"TB_WINNT_6.1_alder_leak_test",1,unix_
timestamp());
Query OK, 1 row affected (0.69 sec)

Comment 6

[Justin Wood (:Callek)]

Attachment: [config] set tinderbox_tree

Sooo, we need to set tinderbox_tree in order to set the --branch properly for graph server.

I know I know, this is *sooo* wrong, but its accurate.

Comment 7

[Armen [:armenzg]]

Comment on attachment 8341852 [details] [diff] [review]
[config] set tinderbox_tree

Review of attachment 8341852 [details] [diff] [review]:
-----------------------------------------------------------------

I thought we had something to deal with this.
However, I would not be sure if it would apply for thunderbird_project_branches.py

Comment 8

[Justin Wood (:Callek)]

Comment on attachment 8341852 [details] [diff] [review]
[config] set tinderbox_tree

checked in and merged to production

Comment 9

[Justin Wood (:Callek)]

Attachment: [config] add test support for TB project branches

This one turned out to be quite easy.

Comment 10

[John Hopkins (:jhopkins)]

Comment on attachment 8359590 [details] [diff] [review]
[config] add test support for TB project branches

This condition isn't needed:

 +            if platform in PLATFORMS:

I added an "assert platform in PLATFORMS" locally and it passed test-masters.sh.
r+ with that condition removed.

Comment 11

[Justin Wood (:Callek)]

https://hg.mozilla.org/build/buildbot-configs/rev/a75b9d3b27cd

And then ran fabric to make sure the symlink existed for all test masters.

Comment 12

[[e-mail] (:bhearsum)]

something is in production