On Wed, 2022-02-02 at 18:06 -0500, Denys Dmytriyenko wrote: > https://git.openembedded.org/openembedded-core/commit/?id=f24c2bf93a9549ae5a21ce378a7bf47ef052ad65 > > I realize these are in master-next and haven't been merged to master yet. > Richard mentioned there are some issues with one of the autobuilder nodes > and hence master-next is taking longer to test lately.
There have been a few issues: a) there is a nasty reproducibility failure which highlighted a hash equivalence bug which is blocking one patch. It took a while to work out the cause though and know which patch was causing the issues. Joshua and I have talked about a fix on irc and one the team call on Tuesday, it isn't an easy one. b) The Centos Stream workers pulled in a problem change which broke builds. When that number of builds start failing, we can't trust the test run results which makes it hard to know which patches to merge. c) I've been unwell over the last few days and you've noticed my productivity dip, I wasn't fixing the autobuilder issues or the hash issues as quickly as I may have done usually. Similarly patch review slowed and so on. I did prioritise meetings, status reports and so on but had limited energy to do everything. For a), I'll send out some patches I have which seem to be working. Once a) was understood, for b) I could merge some patches by multiple runs and averaging but that is slow. We'll drop the number of stream workers we have in the autobuilder as I'm not happy with the issues introduced and the lack of response to fixing them. There is a make hang issue on stream that is unresolved since November and we may drop all the workers if we can't figure that out, not sure yet. Alex did help debug some of this yesterday which has unblocked us (thanks). For c), I'm hopefully recovering. It is a warning/reminder to people that we do need other people stepping up and helping. I do my best but I am one person and I can't change that. > I don't believe bombarding Richard with duplicate patches improves his > productivity. :) On his behalf I'm asking for a bit more patience and > understanding... Thanks, I know you've my interests at heart here :). Alex's series is a little different in that he is rounding up the remaining AUH output and it can overlap with what people send due to the review and testing timescales. I do prioritise anything from others over what Alex has in a series like this, unless there is some reason to do otherwise. Alex and I both have agreed that but it probably isn't obvious on the list. I do sometimes wish he'd rebase it before sending but that then doesn't represent what he tested so there are pros and cons to that and skipping overlapping patches isn't a huge deal. I can/will drop anything that doesn't apply and let Alex sort it ;-). Usually we don't see as much of this as things run faster. As things slowed down it did make it more visible. Cheers, Richard
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#161271): https://lists.openembedded.org/g/openembedded-core/message/161271 Mute This Topic: https://lists.openembedded.org/mt/88865277/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
