545895 - create new linux and win32 ix ref machines

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[[e-mail] (:bhearsum)]

We need to set aside two ix machines to become permanent ref images for our ix farm. We should clone existing images to do this, and then update them with whatever we need to put on them, including fixes from bugs 541953 and 545801.

Comment 1

[[e-mail] (:bhearsum)]

Attachment: opsi key generator, with ix support

Simple update of this script to support the IX machines.

Comment 2

[Chris AtLee [:catlee]]

Comment on attachment 429132 [details] [diff] [review]
opsi key generator, with ix support

>def log(msg):
>    print '%s - %s' % (datetime.strftime(datetime.now(), "%Y-%M-%d %H:%M"), msg)
[snip]
>    # Set up logging
>    try:
>        log = open(options.logfile, "a")
>        sys.stdout = log
>        def cleanup(log):
>            log.flush()
>            log.close()
>        atexit.register(cleanup, log)
>    except IOError:
>        log("WARN: Couldn't open %s, logging to STDOUT instead")

Depending on if you can open the logfile or not, 'log' will either be the function, or the open file.  Can you rename the 'log' variable here to something else?

Comment 3

[[e-mail] (:bhearsum)]

Attachment: fix linux ix machine to be .b.m.o

I had phong update DNS and the machine itself, so Puppet needs updating, too.

Comment 4

[[e-mail] (:bhearsum)]

Attachment: ix support in host key generator, the right patch

Comment 5

[[e-mail] (:bhearsum)]

Comment on attachment 429185 [details] [diff] [review]
fix linux ix machine to be .b.m.o

changeset:   110:4f3b55768ade

Comment 6

[Chris AtLee [:catlee]]

Comment on attachment 429186 [details] [diff] [review]
ix support in host key generator, the right patch

Looks good.  Tiny nit, can you put 'build-ix' on a new line?

Comment 7

[[e-mail] (:bhearsum)]

Comment on attachment 429186 [details] [diff] [review]
ix support in host key generator, the right patch

changeset:   516:b0391e69d8dc

Comment 8

[[e-mail] (:bhearsum)]

I've fixed all the issues noted in bug 541953 on the windows machine.
I've fixed all the issues in bug 545801 on the linux one, as well as installed the PAE kernel.

An IT bug has been filed to pull new images from these. We're all done after that.

Comment 9

[[e-mail] (:bhearsum)]

Windows image is done. Linux image needs grub re-installed and possibly fixes for bugs 549722 and 549672

Comment 10

[[e-mail] (:bhearsum)]

Just updated linux-ix-ref, we're all done here.

Comment 11

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

Just had to repush the /3GB switch in boot.ini to mw32-ix-slave19 after a PGO build failed there. I think what happened is that win32-ix-ref got recreated without that fix, and was used to re-image 19 (bug 545876). I've now repushed the 3GB package to the ref image but I'm wondering if we could be missing any other packages ? Particularly ones that are only setup once.

Do we need to take a new image to 'capture' the fix I pushed ?

Comment 12

[[e-mail] (:bhearsum)]

(In reply to comment #11)
> Just had to repush the /3GB switch in boot.ini to mw32-ix-slave19 after a PGO
> build failed there. I think what happened is that win32-ix-ref got recreated
> without that fix, and was used to re-image 19 (bug 545876). I've now repushed
> the 3GB package to the ref image but I'm wondering if we could be missing any
> other packages ? Particularly ones that are only setup once.
> 
> Do we need to take a new image to 'capture' the fix I pushed ?

The ref image does indeed have the /3gb flag set, apparently on feb 16th. It doesn't look that way, but it's *possible* that slave19 was imaged before that happened.

I had a look around mw32-ix-slave19 and AFAICT all of the things it claims are installed, are truly installed.

And to answer your last question, no, we don't need to capture the fix anywhere. According to IT, we image ix machines directly from the ref ones, not through an image.