paul.larson at linaro.org
Fri Feb 10 16:13:41 UTC 2012
On Fri, Feb 10, 2012 at 6:16 AM, Benjamin Gaignard
> last time I have ask the cable was on panda 03 :-)
> Anyway e2eaudiotest itself isn't mature enough, I believe that we still
> have issue on audio paths settings, but Kurt will work on that (Tom can you
> confirm ?)
I have submit lava e2e test to check if the parser was working well even
> on error cases.
That could be, we need to add tags to the boards that have it so that you
can be sure, but for the audio cables, they are cheap enough that we should
really just put them everywhere. And yes, even if the the test fails
currently, we should still run it if it's ready to go. Is it ready to go?
If so, I'd like to get it merged, and either in the daily runs or in some
special multimedia run if that is preferred.
About cmatest: I have try to test the latest version (v21) but it failed
> http://validation.linaro.org/lava-server/scheduler/job/11852 with this
> strange error "ERROR: There is no test with the specified ID".
> I don't understand where is the problem, I have run the test on my side
> without any issues. How the test ID could be wrong ? The same test ID has
> been used to install the test itself .... strange.
We'll try to look into this today.
> The good news is that realvideo test is working well: http://validation.linaro.org/lava-server/scheduler/job/1;a=blob_plain;f=cma.json;hb=
> HEAD1619 <http://validation.linaro.org/lava-server/scheduler/job/11619>
Great! It's done in json, so we can't merge it, but it would be possible to
get into regular runs anyway. Looks like it takes maybe 45 min. to run.
Any special requirements here?
> I have lot of meetings this morning, and I leave at 12:30....
> What I wanted to talk with you and Tom, is the numbers of automatic tests
> done daily on lava. They take so much time than I'm not able to
> run/validate my own tests before 3-4 pm.
> Do you think you can reduce the number of daily tests ? for example is it
> really needed to do android and ubuntu lab health test daily on each board ?
> Maybe we can also reduce lava test time setup by having rootfs with
> pre-installed lava-tool package.
> Have dedicated lava rootfs could also help me the speed up test creation
> because it is always difficult to found the correct match between hwpack
> and rootfs.
I'm not sure what you mean here, you don't necessarily have to have the
rootfs from the exact same day. As for the backload on the boards and not
getting your job in quickly, there are a couple of things to be aware of:
First, we had an issue with the scheduler earlier in the week and needed to
shut it down for a bit. Short story is that it's back up now, processing
jobs, and it's mostly cleared the queue that had built up. There are now
lots of boards sitting idle (even snowball). Next, on snowball, we only
have 3 at the moment. We're supposed to get some more, but that means for
now they are going to be pretty busy.
The health checks are part of an effort to try to detect when there's
something wrong with the board, or infrastructure that would prevent jobs
from passing when they normally should. They aren't likely to go away, but
we just merged something that should help us make them faster.
> 2012/2/10 Paul Larson <paul.larson at linaro.org>
>> Let's talk on Friday, sorry I missed today. I didn't forget you though,
>> just got pulled into an unexpected meeting late in the day after my tsc
>> talk. It never fails that at connects I never manage to talk to everyone I
>> want to, but I wanted to definitely sync with the two of you and see how
>> things were going with cmatest and e2eaudio.
>> One quick point - on the e2e audio tests, I noticed that you seem to have
>> got a failed result (but a result none-the-less) on panda03. I *think* the
>> only board we've managed to connect with the loopback audio cable is
>> panda01. Dave can confirm that. The company we ordered the audio cables
>> from delayed and we didn't get them before the connect. :(
>> In the meantime, if you set the target for the job to look at panda01,
>> you should be able to get to a board with the loopback actually plugged in.
>> Paul Larson
> Benjamin Gaignard
> Multimedia Working Group
> Linaro.org <http://www.linaro.org/>* **│ *Open source software for ARM
> Follow *Linaro: *Facebook <http://www.facebook.com/pages/Linaro> | Twitter<http://twitter.com/#%21/linaroorg>
> | Blog <http://www.linaro.org/linaro-blog/>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the linaro-validation