Hey Guys,
We currently use ubuntu-desktop RFS's for our daily health checks in LAVA. Its my understanding the dev-platform team will be sunsetting ubuntu-desktop images.
Due to this, I think we (the LAVA team) need to think about moving to a new set of images for our health checks so that they better reflect reality. 2012.11 should be producing some pre-built images for both server and nano. I think picking the server pre-built images probably makes more sense since it gives us a little more coverage than nano, but I don't have a strong sense on whether it makes much difference since we basically just do boot-testing in our health check.
= So what does this mean?
I think we'll need to file a 12.12 blueprint for doing health check investigation work. Essentially, we should build a job for each device-type in production based off its health job. We then update the image URL to be the new candidate. We then submit it 100 times and do a failure analysis. At this point we cross our fingers and hope the failures aren't worse than what we currently see. If so, we can switch the image over. If its not the case, we'll need to work with dev-platform team on getting new issues addressed.
make sense? who wants to help? :)
-andy
Andy Doan andy.doan@linaro.org writes:
Hey Guys,
We currently use ubuntu-desktop RFS's for our daily health checks in LAVA. Its my understanding the dev-platform team will be sunsetting ubuntu-desktop images.
Due to this, I think we (the LAVA team) need to think about moving to a new set of images for our health checks so that they better reflect reality. 2012.11 should be producing some pre-built images for both server and nano. I think picking the server pre-built images probably makes more sense since it gives us a little more coverage than nano, but I don't have a strong sense on whether it makes much difference since we basically just do boot-testing in our health check.
I'd go for nano for the simple reason that they are smaller and so deploy quicker, at least until/unless we start actually running some tests in the health jobs.
Server and nano images might well use different kernels (LT vs upstream) so we may want to just use whichever one boots better.
= So what does this mean?
I think we'll need to file a 12.12 blueprint for doing health check investigation work. Essentially, we should build a job for each device-type in production based off its health job. We then update the image URL to be the new candidate. We then submit it 100 times and do a failure analysis. At this point we cross our fingers and hope the failures aren't worse than what we currently see. If so, we can switch the image over. If its not the case, we'll need to work with dev-platform team on getting new issues addressed.
make sense?
Yeah.
who wants to help? :)
Not me :)
Cheers, mwh
On Monday 19 November 2012 10:33 PM, Andy Doan wrote:
I think we'll need to file a 12.12 blueprint for doing health check investigation work. Essentially, we should build a job for each device-type in production based off its health job. We then update the image URL to be the new candidate. We then submit it 100 times and do a failure analysis. At this point we cross our fingers and hope the failures aren't worse than what we currently see. If so, we can switch the image over. If its not the case, we'll need to work with dev-platform team on getting new issues addressed.
make sense? who wants to help? :)
Though I did not get the complete idea of what has to be done, I would like to help in doing this :) Please point me to the blueprint.
Thank You.
On 19 Nov 2012, at 17:03, Andy Doan andy.doan@linaro.org wrote:
Hey Guys,
We currently use ubuntu-desktop RFS's for our daily health checks in LAVA. Its my understanding the dev-platform team will be sunsetting ubuntu-desktop images.
Due to this, I think we (the LAVA team) need to think about moving to a new set of images for our health checks so that they better reflect reality. 2012.11 should be producing some pre-built images for both server and nano. I think picking the server pre-built images probably makes more sense since it gives us a little more coverage than nano, but I don't have a strong sense on whether it makes much difference since we basically just do boot-testing in our health check.
= So what does this mean?
I think we'll need to file a 12.12 blueprint for doing health check investigation work. Essentially, we should build a job for each device-type in production based off its health job. We then update the image URL to be the new candidate. We then submit it 100 times and do a failure analysis. At this point we cross our fingers and hope the failures aren't worse than what we currently see. If so, we can switch the image over. If its not the case, we'll need to work with dev-platform team on getting new issues addressed.
make sense? who wants to help? :)
I've probably got most experience in this, so put me down for it. :)
Dave
-andy
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
On Tuesday 20 November 2012 11:51 AM, Dave Pigott wrote:
= So what does this mean?
I think we'll need to file a 12.12 blueprint for doing health check investigation work. Essentially, we should build a job for each device-type in production based off its health job. We then update the image URL to be the new candidate. We then submit it 100 times and do a failure analysis. At this point we cross our fingers and hope the failures aren't worse than what we currently see. If so, we can switch the image over. If its not the case, we'll need to work with dev-platform team on getting new issues addressed.
make sense? who wants to help? :)
I've probably got most experience in this, so put me down for it. :)
Then I will help you :)
Thank You.
On 20 Nov 2012, at 06:22, Senthil Kumaran S senthil.kumaran@linaro.org wrote:
On Tuesday 20 November 2012 11:51 AM, Dave Pigott wrote:
= So what does this mean?
I think we'll need to file a 12.12 blueprint for doing health check investigation work. Essentially, we should build a job for each device-type in production based off its health job. We then update the image URL to be the new candidate. We then submit it 100 times and do a failure analysis. At this point we cross our fingers and hope the failures aren't worse than what we currently see. If so, we can switch the image over. If its not the case, we'll need to work with dev-platform team on getting new issues addressed.
make sense? who wants to help? :)
I've probably got most experience in this, so put me down for it. :)
Then I will help you :)
Hi Senthil,
Sorry - I jumped in before seeing your reply. OK. We can work it together. Given our closeness in timezones, it will probably work quite well.
Thanks
Dave
Thank You.
Senthil Kumaran S http://www.stylesen.org/ http://www.sasenthilkumaran.com/
linaro-validation@lists.linaro.org