Hey Ben - 

tl;dr - Inaris won't go anywhere unless there's an exigent reason to :-) And 
let us know how/if our coverage balance is meeting needs! 

As part of a larger effort, and as you already know from below, we're rapidly 
moving off Unagis (see also 
https://bugzilla.mozilla.org/show_bug.cgi?id=924503). Our community (in 
particular, but not solely -- third-party contracts, too) won't (always) have 
access to Hamachi, Leos, or Helix devices, though, so we have actually retained 
support for Inaris, and their unlocked bootloaders and RelEng builds. 

We'll always continue to re-evaluate the mix of devices/branches (and the 
combinations thereof), matched with our limited ability (few folks in Mountain 
View, as it stands right now) to keep the devices maintained (Hamachis are 
turning out to be harder to get the flashing/perf-scripts for, than were/are 
Inaris and Unagis). 

To specifically answer your question, "Do we have an alternative method for 
detecting hardware specific bugs?" - both the Gaia UI Tests [1] and the 
performance tests (perf, mozperf + fps [2]) can be run locally, so if the team 
has the particular device, and the appropriate build from RelEng is still 
available (or you can self-build one), you shouldn't be /too/ stuck. 

(As part of a larger discussion, going forward, it'd be great as a larger team 
to (re?)define what "supported" means in terms of builds, job types (UI/perf) 
for each branch, and each device.) 

Until now, I wasn't aware of a device-specific issue on the Inari (esp. wrt 
perf), so am happy to maintain a couple Inari nodes, as Dave suggests, as long 
as they're continuing to provide value, and are supported by 
RelEng/product/devs et al. Feedback like that (that our setup can help 
troubleshoot or baseline a metric) is very helpful in knowing how important and 
valuable the coverage we help maintain is (compared with the rest of the 
branches/devices), so thanks! 

Esp. as you and the others on the Performance team have needs like this, don't 
hesitate to reach out to us (email/IRC), so we can work together to figure out 
what sustainable coverage looks like - it's never our intention to drop 
something that's useful; we do so most often to free up capacity for something 
else :-) 

Thanks, 

- Stephen 

[1] http://qa-selenium.mv.mozilla.com:8080/view/B2G%20Hamachi/ 
[2] http://qa-selenium.mv.mozilla.com:8080/view/B2G%20Perf/ 

----- Original Message -----

From: "Ben Kelly" <[email protected]> 
To: "Dave Hunt" <[email protected]>, [email protected] 
Cc: "Stephen Donner" <[email protected]> 
Sent: Tuesday, October 29, 2013 10:15:46 AM 
Subject: Re: [b2g] Stopping performance testing on Unagi 

On 10/29/2013 12:45 PM, Dave Hunt wrote: 


If the results we're gathering are proving useful then I would say we 
should continue to keep a couple of these devices attached and running 
the tests for now. We can always revisit this again when we have a need 
for testing another device, but I would say we're unlikely to want to 
maintain running tests for several different devices at once. 

Do we have an alternative method for detecting hardware specific bugs? 




I would think over time we would be adding more devices instead of 
unifying to a single device. I could see a core device where we run 
tests frequently and then secondary devices that run perhaps once a day 
or something, but I wouldn't think we would want to remove those devices 
completely. 

Thanks! 

Ben 


<blockquote>

Dave 

On 10/29/13 4:31 PM, Ben Kelly wrote: 

<blockquote>
On 10/18/2013 10:04 AM, Ben Kelly wrote: 

<blockquote>
On 10/18/2013 04:15 AM, Dave Hunt wrote: 

<blockquote>
This is a notification that the performance testing on Unagi (links 
below) was stopped this week in favour of Hamachi and Inari. Once we're 
satisfied with the reliability of Hamachi we intent to drop Inari too. 
Please let me know if you have any concerns. 

</blockquote>

Thanks for the heads up Dave! 

Just out of curiousity, why are we dropping inari? Isn't that also a 
shipping platform we need to support? 

</blockquote>

For example, bocoup was asking today in the gaia meeting about scroll 
performance on inari that doesn't seem to effect the other platforms. 
If we're only measuring hamachi, I'm worried we're going to miss this 
kind of stuff. 

Thanks again. 

Ben 

</blockquote>

-- 
*Dave Hunt* 
QA, Mozilla Corporation 
[email protected] 


</blockquote>
_______________________________________________
dev-b2g mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-b2g

Reply via email to