629690 - slave-side slave-alloc support for win64

Status: RESOLVED FIXED

(Release Engineering :: General, defect, P4)

Description

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

+++ This bug was initially created as a clone of Bug #616351 +++

Similar to bug 616003, but for win64 systems.  These systems are as yet not in production, so doing the deployment now will be easier than doing it later!

Comment 2

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

Not blocking bug 605278 anymore, as these systems aren't in production.

Comment 3

[Armen [:armenzg]]

Initial steps:

cd /c/
wget http://hg.mozilla.org/build/puppet-manifests/raw-file/d55a0757f0f0/modules/buildslave/files/runslave.py
cd /c/Users/cltbld/AppData/Roaming/Microsoft/Windows/Start\ Menu/Programs/Startup
wget -Obuildbot.bat http://hg.mozilla.org/build/opsi-package-sources/raw-file/2d51f32ff0f2/buildbot-startup/CLIENT_DATA/buildbot.bat
cd /d/mozilla-build
wget -Ostart-buildbot.bat https://bug565402.bugzilla.mozilla.org/attachment.cgi?id=537611

Comment 4

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

Check that runslave.py is really in C:\ - the VirtualVault stuff messed me up.

Comment 5

[Armen [:armenzg]]

Attachment: newer mozilla-build installs python on new path

Reviewed by dustin on IRC
http://hg.mozilla.org/build/puppet-manifests/rev/a43555c557c6

Comment 6

[Armen [:armenzg]]

This got added on https://wiki.mozilla.org/ReferencePlatforms/Win64#Startup_scripts_.26_slave-alloc

Waiting on machines to be re-imaged so we can test it works for any other host besides the ref machine.

We could try it on mw64-ix-slave01 but putting on the side for now.

Comment 7

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

In the interim, what other testing can we do on the new win64 image?  Is it possible to get SSH installed, or tweak the VNC settings so they're better than the W764 talos machines were?  Time sync?  Cleanup tasks similar to those run on the w32 builders?

Comment 8

[Armen [:armenzg]]

It is all installed. It even has the driver that makes VNC to work better. Please check the ref docs if you want to double check anything.

The only part that is missing is to add entries to the actual slave alloc DB and verify that slaves connect to the specified masters.

Comment 9

[Armen [:armenzg]]

Attachment: ensure that basedir exists

Comment 10

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

Comment on attachment 544794 [details] [diff] [review]
ensure that basedir exists

I'm not sure why the existing basedir-creation code doesn't work, but it would be much better to make this directory when writing the buildbot.tac (in the download method), rather than in a "get_" function (which should not have any side-effects).

Comment 11

[Armen [:armenzg]]

Attachment: ensure that basedir exists v1

Comment 12

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

Comment on attachment 544831 [details] [diff] [review]
ensure that basedir exists v1

Review of attachment 544831 [details] [diff] [review]:
-----------------------------------------------------------------

::: modules/buildslave/files/runslave.py
@@ +118,5 @@
>              elif slave_matches('xp', 'w7', 'w764'):
>                  basedir = dirs['win_test']
>  
> +        # ensure the basedir exists so we can write the buildbot.tac file
> +        self.ensure_basedir_exists()

This call should go away - this method should only be called just before creating the tac file

Comment 13

[Armen [:armenzg]]

Attachment: ensure that basedir exists v1

Addressed the comment. r+ through IRC.

Comment 14

[Armen [:armenzg]]

Comment on attachment 544834 [details] [diff] [review]
ensure that basedir exists v1

This got landed and deployed:
http://hg.mozilla.org/build/puppet-manifests/rev/1bbb72f7dcea