776987 - Set up one buildbot master and a foopy on SCL1 for acceptance testing

Status: RESOLVED FIXED

(Infrastructure & Operations :: RelOps: General, task)

Description

[Armen [:armenzg]]

We will need a buildbot master to handle the initial set of pandas.
A foppy is also needed. I believe this can be a Linux foopy since Callek got the work most of the way done.
I will let Callek follow up on spec for the foopy.

We will add more buildbot masters and foopies once we're further down the line of purchasing.

This should do to test the final setup (foopy spec is still to be settle AFAIK).

Thanks in advance!

Comment 1

[Justin Wood (:Callek)]

(In reply to Armen Zambrano G. [:armenzg] - Release Engineer from comment #0)
> We will need a buildbot master to handle the initial set of pandas.
> A foppy is also needed. I believe this can be a Linux foopy since Callek got
> the work most of the way done.
> I will let Callek follow up on spec for the foopy.

CentOS6.2 kickstarted for PuppetAgain. We'll need to add a node-name, etc.

Comment 2

[Amy Rich [:arr] [:arich]]

Do we have any preference on hostnames for foopies (do we even still want to call them foopies, since this seems like a good time to break that naming convention if not)?

Comment 3

[Amy Rich [:arr] [:arich]]

I've created buildbot-master29.build.scl1.mozilla.com for this purpose.  I'm not sure who on the releng side is setting it up.

Comment 4

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

mobile-controllerNN

Comment 5

[Armen [:armenzg]]

dustin's proposal works. I would prefer not to have a new name since I really hope we can get rid of the foopy model all together in the next 2-3 quarters; either way works. I can also see how a different spec might need for a new name but could get on the way of our other tools.

I will let Callek have the last say (since he deals with them much more).

Comment 6

[Armen [:armenzg]]

(In reply to Amy Rich [:arich] [:arr] from comment #3)
> I've created buildbot-master29.build.scl1.mozilla.com for this purpose.  I'm
> not sure who on the releng side is setting it up.

Filed as bug 777370.

Comment 7

[Justin Wood (:Callek)]

(In reply to Armen Zambrano G. [:armenzg] - Release Engineer from comment #5)
> dustin's proposal works. I would prefer not to have a new name since I
> really hope we can get rid of the foopy model all together in the next 2-3
> quarters; either way works. I can also see how a different spec might need
> for a new name but could get on the way of our other tools.
> 
> I will let Callek have the last say (since he deals with them much more).

Short term, I want foopyNN

Much of our current code, and dashboard code relies on this naming pattern. I *really* want to move away from that requirement and plan to, but I also don't want to have to refactor tons of additional code that can be solved short-term with just naming.

Comment 8

[Amy Rich [:arr] [:arich]]

foopy25 and up it is.  As soon as one of the hps is freed up, I can build the one in scl1 for the pandas.

Comment 9

[Amy Rich [:arr] [:arich]]

foopy25.build.scl1.mozilla.com has been kickstarted with the puppetized centos 6.2 image.

Comment 10

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

Could you add the .build.mozilla.org CNAME to DNS ?

$ host buildbot-master29.build.mozilla.org 10.2.74.125
Using domain server:
Name: 10.2.74.125
Address: 10.2.74.125#53
Aliases: 

Host buildbot-master29.build.mozilla.org not found: 3(NXDOMAIN)

Comment 11

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

It was pointing to mtv1.  Fixed.

Comment 12

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

Thanks.