Closed
Bug 737282
Opened 14 years ago
Closed 14 years ago
Image a rev 3 mini with win7 32bit for PDF.js acceptance tests
Categories
(Infrastructure & Operations :: RelOps: General, task)
Tracking
(Not tracked)
RESOLVED
INVALID
People
(Reporter: cmtalbert, Assigned: dividehex)
Details
= tl;dr =
I have an old rev3 mac mini that I would like to get imaged to win7 32bit.
= Details =
I have a rev3 mini left over from previous automation that is no longer needed. I'd like to give it to the PDF.js team to use as their pre-merge test automation.
Here's how the automation solution will work.
1. PDF.js is an extension that lives in github. When someone submits a pull request to the github, the PDF.js team causes that to run their reftests.
2. These reftests run on a system (ideally that system is this mini)
3. Once the PDF.js team has a new version of their extension, they merge to mozilla-central.
So, to make the merge in step 3 as painless as possible, the system they run those reftests on in step 2 should be as close to the hardware we are running in production as possible. They are mostly concerned about windows support, since hardware acceleration of windows is pretty much on-everywhere by default.
If we get windows working, might extend this solution later for mac testing (using a newer, more available mac mini).
The mac mini is the one on my standup desk beneath the A*team poster that has a yellow post-it on it saying "PDF.JS".
Comment 1•14 years ago
|
||
Hi, clint, what's the priority on this? Unless it's tied to a high priority q1 goal, we likely won't get to it for a few weeks.
Comment 2•14 years ago
|
||
We *really* should not be putting more minis into service for random tasks like this. Minis are truly an abomination and a HUGE time-sink for us -- the old ones more so. Can this run in a VM?
(In reply to Dustin J. Mitchell [:dustin] from comment #2)
> We *really* should not be putting more minis into service for random tasks
> like this. Minis are truly an abomination and a HUGE time-sink for us --
> the old ones more so. Can this run in a VM?
I heartily agree with you. However, the PDF system uses hardware acceleration to display its graphics (which is what this would be testing). That means that if we are to use a VM (that's what they currently do - they use an EC2 VM) then there is a very large chance that the pre-merge test will pass and then when we actually merge this into mozilla-central the tests will fail due to hardware rendering differences that we could (and should) have caught earlier.
I would love for us to move away from mac mini's wholesale for windows and linux testers. But that's a larger issue that's been going on for over two years now. Current work on that appears stalled in bug 691856.
Arr: this isn't a Q1 goal. We can do this in Q2.
Comment 4•14 years ago
|
||
(In reply to Dustin J. Mitchell [:dustin] from comment #2)
> We *really* should not be putting more minis into service for random tasks
> like this. Minis are truly an abomination and a HUGE time-sink for us --
> the old ones more so. Can this run in a VM?
I'm not sure Clint is advocating putting this single mini into the existing automation, he simply wants it re-imaged with a different OS so he can use it as required for PDF.js. Please correct me if I'm wrong.
(In reply to Clint Talbert ( :ctalbert ) from comment #3)
> I would love for us to move away from mac mini's wholesale for windows and
> linux testers. But that's a larger issue that's been going on for over two
> years now. Current work on that appears stalled in bug 691856.
Yes, that work is stalled only in that there are a limited number of releng people that could work on it, and the ones that can are currently busy doing the critical path work for the in-progress colo move by IT. I'll follow-up in bug 691856 with more details.
(In reply to Chris Cooper [:coop] from comment #4)
> (In reply to Dustin J. Mitchell [:dustin] from comment #2)
> > We *really* should not be putting more minis into service for random tasks
> > like this. Minis are truly an abomination and a HUGE time-sink for us --
> > the old ones more so. Can this run in a VM?
>
> I'm not sure Clint is advocating putting this single mini into the existing
> automation, he simply wants it re-imaged with a different OS so he can use
> it as required for PDF.js. Please correct me if I'm wrong.
>
That's exactly correct.
> (In reply to Clint Talbert ( :ctalbert ) from comment #3)
> > I would love for us to move away from mac mini's wholesale for windows and
> > linux testers. But that's a larger issue that's been going on for over two
> > years now. Current work on that appears stalled in bug 691856.
>
> Yes, that work is stalled only in that there are a limited number of releng
> people that could work on it, and the ones that can are currently busy doing
> the critical path work for the in-progress colo move by IT. I'll follow-up
> in bug 691856 with more details.
Ah cool. Thanks Coop.
Comment 6•14 years ago
|
||
Clint, please put these minis on Jake's desk.
Assignee: server-ops-releng → jwatkins
colo-trip: --- → mtv1
Thanks Amy and Jake. The mini is over there. I added bug numbers to my post-its to make things clear. Let me know if you need anything else.
| Assignee | ||
Comment 8•14 years ago
|
||
Clint: we were unable to image this mini since it is actually an r2. The only win7 image we have was built on the r3 hardware. I gave it a shot but as suspected it won't work on an r2.
| Assignee | ||
Comment 9•14 years ago
|
||
Clint: I will leave this mini in AFK for you
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → INVALID
Updated•13 years ago
|
Component: Server Operations: RelEng → RelOps
Product: mozilla.org → Infrastructure & Operations
You need to log in
before you can comment on or make changes to this bug.
Description
•