616351 - slave-side slave-alloc support for win32

Status: RESOLVED FIXED

(Release Engineering :: General, defect, P3)

Description

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

Similar to bug 616003, but for win32 systems.  As I understand it, these are build slaves only, and are controlled by OPSI, so this means deploying runslave.py with OPSI and adjusting the other equipment to run it.

Comment 1

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

I can do this, but this would be a great place to get some help on this project (along with bug 629690 and bug 629692)

Comment 2

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

Attachment: m616351-opsi-package-sources-r1.patch

This is the OPSI config to install runslave.py.

This drops both of the start-buildbot-repeatedly loops.  I'll run this in staging for a while, and re-add that (well, one loop anyway) if it turns out to be necessary.

I had trouble with bash crashing while running the shell script (yes, really), so I took that out.

This works and runs runslave.py, but the currrent 'tip' version of runslave has a typo that prevents it from starting things up.

Comment 3

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

Attachment: m616351-puppet-manifests-r1.patch

This fixes the typo in runslave.py (indentation error that caused it not to find python.exe under D:).

Comment 4

[[e-mail] (:bhearsum)]

Comment on attachment 533168 [details] [diff] [review]
m616351-opsi-package-sources-r1.patch

Looks reasonable to me.

Comment 5

[Mike Taylor [:bear]]

Comment on attachment 533169 [details] [diff] [review]
m616351-puppet-manifests-r1.patch

*stamp*

Comment 6

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

Attachment: m616351-opsi-package-sources-p2.patch

Oops, I don't want the package to always install 'tip' - rather, it should specify a particular hg revision, and we'll bump the package revision when that changes.

Comment 7

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

Comment on attachment 533370 [details] [diff] [review]
m616351-opsi-package-sources-p2.patch

I'm seeing problems with the startup failures - it looks like some percentage of the invocations of Python just fail immediately.  This used to happen, so it's not horribly surprising that it's still going on.  I'll try to write a loop that will restart runslave.py if it fails quickly, but will just die if it takes a while to exit (e.g., after a build).

The staging slaves are also failing to build with an incorrect %PATH%.  That will need to be fixed first.

Comment 8

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

Attachment: m616351-opsi-package-sources-p3.patch

This updates the revision to point to the new runslave.py from bug 658024, and fixes the %PATH% problem.  It doesn't address the looping problem because I haven't been able to replicate it since the first time I ran runslave.py (and that may have been operator error).

Comment 9

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

Comment on attachment 533469 [details] [diff] [review]
m616351-opsi-package-sources-p3.patch

Still working on this - bug 658263 - before pushing it into production.

Comment 10

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

OK, deployed to production.