No Android 3.18 test builds for 2 days about kernelci-core

[broonie]

We're experiencing serious stability problems with the Jenkins master and have disabled a lot of builds of trees that aren't mainline focused in order to reduce load on it and avoid triggering the issues, yours unfortunately got caught up in this.

We're in the process of redeploying to new instances which should address the capacity issues but unfortunately the Jenkins master is one of the last bits to be moved over.

[lag-linaro]

It's a shame since I have made KernelCI a critical piece of my workflow.

Without it I'm forced to push backports to android-3.18 lacking full testing exposure.

FYI: android-3.18 is consumed by real/deployed devices.

Stable is as important as Mainline in this regard.

Any way this branch can be 'turned back on'?

[gctucker]

This branch isn't disabled, only that random jobs fail because of an issue between Jenkins and Docker on the remote builders. Also the kernelci.org server has just been replaced with a new one, this isn't Jenkins but it took some time to set it up. Fixing Jenkins it the next item on the list.

@lag-linaro kernelci.org is not about testing any product in particular but the upstream Linux kernel. Just like v3.18 was the latest mainline a few years ago, we are now testing new kernels that will get deployed on future products. That is the primary goal of kernelci.org as a service. However, it is possible to also test downstream branches with the same KernelCI tools. Because Android kernels have a precedent of being tested on kernelci.org for historical reasons, we should discuss how to proceed and settle this conflict. There are many other similar use-cases, such as distribution kernels or various very-long-term-stable branches maintained by third parties. We can't keep having these branches interfere with testing mainline, next and all upstream-oriented branches on kernelci.org directly but there are ways to accommodate them (priorities, pools of builders, separate configurations...). Let's plan a meeting or email discussion about that.

[lag-linaro]

Seems to be working well again. Thank you.

Closing this for now.

[gctucker]

@lag-linaro Well it may seem to be working again, but the issue I described in my previous comment still hasn't been resolved. I guess we can move the discussion to emails.

[lag-linaro]

Happy to chat via email.

I've messaged you in the past, but received no reply.

[lag-linaro]

Something else going on? I've only had 1 build this week.

This really does stop me in my tracks.

[gctucker]

Yes, lots going on with moving servers around in production. The builds were not paused for that long though, so maybe you've been unlucky with the timing somehow. It should all be back on track now, sorry for the inconvenience.

[lag-linaro]

I'll try to push another SHA.

Not sure when the window was, but my builds from late Tuesday and all day yesterday were ignored.

[gctucker]

It looks like there are also some issues with some Jenkins servers, taking a look. That deserves a separate issue though, this one was initially about the 3.18 branch being disabled but it's resolved now as all branches are being monitored.

[lag-linaro]

I didn't want to keep cluttering up your repo with new Issues every time this happens.

Thanks for reporting back though. It makes me feel a bit better that you've found the cause.

[gctucker]

For quick questions, can also get in touch using the #kernelci IRC channel on Freenode.

[lag-linaro]

Wonderful. Will do.