Conversation
Edited 1 year ago
small progress on the kernel testing side project: I did setup LAVA instance in addition to Jenkins and made it publicly available.

it took few weeks due to issues with hardware :(

this is home server so can't provide excellent stability and uptime but here it is:

https://lava.kerneltesting.org
https://jenkins.kerneltesting.org
2
0
2
Edited 1 year ago
Fun part of building a home server with a domain name is the public IP can change. I wrote a script updating DNS A record periodically because of this. This is not a perfect solution though; DNS propagation time takes some time. But it should be sufficient for my service.
1
0
1
@vbabka btw how do you think about maintaining a slab/experimental branch so that it can be tested first even before being reviewed and went to slab/for-next?
1
0
0
@ptesarik nope ;) some DNS hostimg services provide updating records via web APIs!
1
0
1

@hyeyoo Oh, okay. Providers could still implement the web API call as a RFC2136 dynamic update in the backend, but I guess they don't bother.

0
0
1
Edited 1 year ago
@vbabka

https://git.kerneltesting.org/slab-experimental

Here it is!

This is the first kernel repository that adopts "merge first and then drop if it's broken" policy ;)
1
0
2
@hyeyoo oh great, thanks a lot! You sure it's the first one with that policy, though? Haha.
1
0
1
@vbabka tbh I'm not sure, but at least git repos that are regularly merged to linux-next shouldn't use that policy ;)

Anyway the policy seems to work quite well!

https://lore.kernel.org/linux-mm/CAB=+i9Sw1YSdUKrjygA5cOsVjQMVmS8-KJ+ku4AG9Fw_2guENQ@mail.gmail.com/T/#t
1
0
1
@hyeyoo then why has @ljs broken -next again just days ago 🤭
2
0
2
@vbabka @hyeyoo -next? I thought it was called -fucked?
0
0
2
@vbabka @ljs oh, andrew was the first one with that policy?
2
0
2
@hyeyoo @vbabka it's gotten worse too, I even managed to get my series in that saner-Andrew hadn't merged in the past

mwahahaha
0
0
2
@hyeyoo @ljs not sure if Andrew or the stable tree was the first...
0
0
2