Jenkins Governance Meetings, 1 Jun 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Jenkins Governance Meeting June 1, 2022

Description

Jenkins governance meeting June 1, 2022 with topics including:

* News
* Projects - Java 11
* Projects - Java 17
* Outreach - She Code Africa

A

Cloud, okay, so a very late start, but welcome to the jenkins governance meeting for june 1st. uh We have uh I'm gonna get this wrong again basil and myself, um and I guess we have some news to go over.

B

Yeah so um two areas of news: first, um the 2.346.1 lts. It's been delayed a few weeks to offer some time to address regressions, um but we've got some additional content coming in. We just finished the second round of backboarding today, many thanks to alexander brandes for doing these back ports and we're looking like it's in good shape for 2.346.1.

B

On the weekly side, we've been doing some improvements to the docker images and a lot of ui regression fixes that have been in flight. So many thanks to everyone that has been developing and reviewing and testing regression fixes. That's been truly inspiring to me to see all this community effort going into stabilizing this release.

B

So that's been a flight in the weeklies. um A couple of smaller changes in the weeklies that might otherwise go by and notice are that we finally disabled the and class loader by default, which was a long-standing project from last year, um is that going to affect.

A

End users at all or just internals,.

B

No, it's all it's all internals, although there is a an escape patch in case it causes any unforeseen regressions, but it should uh should not cause any problems as far as we know, so it's purely internal change, but um so there's that and then there's also um the the only other thing that I felt interesting enough to mention was that I noticed the other day we're still mining on an older version of juice and could use some help, upgrading juice to the newest version.

B

People who are regulars to jenkins core development might recall that we were behind on guava for a very long time. So I would hate to see this being year, one of another decade of us being behind on juice.

B

So it would be great if, uh if anyone was interested in trying to complete that upgrade, um that's that's an area that we're starting to fall behind in but other than that the the weeklies are looking great. So.

A

For people like me, who don't know what juice is, what does juice do and how can people help with this? Well, it's the.

B

It's a way of injecting java classes into other java classes. It's a dependency injection system to allow one java class to access an instance of another java class at one time, and it's actually one of many such dependency managers that we have in core. We've also got another one called sas pause.

B

That does a similar thing and I think the um you know the like, like with all of these older parts of jenkins, there's not a lot of knowledge left from the original authors of the code, so this is kind of an area where um being able to read the code and explore and and teach yourself context from the existing comments and from the existing commit history from the existing code and that's going to be a very valuable skill to have- and you know that's an area where there probably won't be too many people left that can help.

B

But once once someone does build up that context, we could then kind of recover from where we left off and make sure that it gets passed down from one person to the next going forward.

B

So there's uh like, like with the ant class loader stuff, there's a lot of historical research there that I think, would be needed to work and that's what makes it a non-trivial um thing.

A

To upgrade yeah there's also testing, I assume, would be a great people who can run there's a docker image for running pr testing. But you know windows never gets enough testing, so anyone who has who can just volunteer sometime to run the builds is probably also very useful.

B

Yeah yeah, in this case the um the automated core pr tests failed on the upgrade. So it was something that we caught very quickly rather than you know doing manual testing, but manual testing would not hurt either for something like that: yeah cool. So for the project side, uh oh, I want to pause it did you want to talk about any news more again.

A

uh I don't have any news for this week. um Nothing comes to mind, I'm a little, not organized, as I usually am so yeah.

B

Okay, so for the on the project side we're getting closer to june 21st, the june 21st weekly, which is when we plan on requiring java 11 and unlike the amp class loader change that will affect end users. um We are looking pretty good for a delivery on that date. um There was a blocker bug that we discovered that has been fixed and back ported, so we're looking we're looking to be in good shape for that from the perspective of core and another small announcement. Is that we're planning on taking java 17 out of preview mode?

B

At the same time, so I will be planning to have full support for java, 11 and ansa, and preliminary support for java 17. upgrading from upgrading java 17 from preview mode into um something that we would feel more confident recommending.

A

That people use so we're not we're not going to recommend anyone use 17 yet, but if they are using 17, it probably should work.

B

Yeah, not only that my my experience has been that 17 is typically more stable than 11 in all the testing that I've done.

B

For example, this blocking bug that we discovered that have to be worked around in pipeline groovy and script security that only affected java 11.. um In fact, what I did.

A

On my dream was.

B

I know I'm sure I got hit by that as well, so yeah yeah. So what? In fact what I did upstream was to backport the change from java 17 down to java 11.. So I think for the most part I would expect 17 to be more stable, but I think 11 has had more testing.

B

So basically my plan is, I think we are recommending that people use java 11. But if you run into a problem it may very well be fixed in 17.

A

So I think that the key one is, we no longer have to tell people not to use 17. right right.

B

The other part of the java 11 work is this jax b removal, so jax b was removed in java 9. I think, and we've had to adapt a lot of plugins that were relying on it. So there's three left that have more than 500 installations that have not been adapted um notification, plug-in I've been working with the maintainer who has already merged the pull request, but is struggling to release it. So hopefully we should get that released soon.

B

I think it's just a matter of resetting the credentials uh with a solo c count, I'm going to adopt that plug-in myself and then merge and release the fix, so that should be taken care of soon as well, then there's hpe plug-in and we got a response from the maintainers and they plan on adding java 11 support by the end of june, which goes well with our dates that we've picked.

B

um So I think we're in good shape for jack speed, compatibility uh for any plug-in with more than 500 installations and for the the plug-ins with fewer than 500 installations.

B

We have submitted pull requests in many cases, but sometimes these pull requests have been ignored. So if these pull requests don't get fixed by september, we're most likely just going to deprecate those plug-ins formally but still give people some still give the maintainers a little bit more time to update them.

B

A

Really all I have for I'm just totally on board with killing old all plugins. They cause a lot of support, both in the plug-in site and deform stuff. So anything that's been long abandoned and not used. I think we should be more aggressively killing yeah and there's. There's.

B

A great example of that recently, with um one of the front-end changes to improve form validation. um We did a code search and discovered that it would break a very old plug-in as part of of improving that form, validation, we're going ahead and deprecating. That plug-in.

B

I don't remember the name of hand down, but there's a lot of examples of that, and I think it's a good pattern of development that we're starting to see where it's all right to evolve core and to leave certain plugins behind on a case-by-case basis, but along with leaving them behind communicating that through a formal deprecation. It's a good way to set users expectations, yeah.

A

And I think we I mean as a group, we also need to push pipelines more because I think a lot of those old plugins are designed because pipeline didn't exist so like there are integrations that are just freestyle in nature that don't make any sense in pipeline and you can work more flexible in pipeline, but they still exist and people use them. So I think we could do a better job of being hey.

A

Yeah learn, learn new things, not you're wrong. I want to stress that freestyle is not wrong, but we could do a better job of being hey. These things are going to make your life easier, yeah.

B

You know, I think some of our some of our users have a lot of jobs running on older job types and it can take a long time to rewrite these older jobs. But it's like you said it's an exercise. That's well worth doing from a supportability perspective, yeah cool! So that's all.

A

B

A

Yeah, so I can cover the other two topics here, not as well as mark could it's kind of I'm kind of I miss mark already, but he's just taking a week off to spend with family. So you can't really, you know, complain um so she she code, africa, uh the contributes on is now complete. uh There were a couple of posts on the like retro type posts on the forum, so people posted you know what they learned.

A

You know what they get, what skills to gain, who they are, that kind of thing, they're very nice to see- um and I guess, as mark wrote, a comment here- he's going to write up a complete blog post for the end or for the like a wrap up. We also summer google, somewhere code. We had uh the four projects um selected they've started to kick off, um there's they're, currently in discussions of essentially where they're going to be doing mostly communication.

A

It sounds like there's a split between um getting more getter channels and a couple slack it's in the cdf slack, um but you know they're moving forward and you know all the mentors are very excited and then for uh for um forms and community topics. I don't really have anything this week uh there was a call out on the forums for uh any help topics. People had so I'm gonna just skim this really quickly, while we're waiting.

B

Shoot that's lost. It.

A

We'll I'll add the I'll add the uh I'm going to add links for she, codes astr for these posts and then there's one.

A

I saw that people are just looking for help yeah. So if anyone has blog posts, uh whoops call out for youtube videos, vlog posts.

A

uh Linux but otherwise it's been fairly quiet on new content. A lot of questions um I'm expecting our june 21st one will be a messy forum stuff, but mostly we just tell people to upgrade and we should be good.

B

Yeah, as usual, we'll upgrade the plugins both before and after upgrading, core yeah.

A

B

That's really the big message.

A

And yeah, so I don't have anything else. This week it's been pretty quiet week.

A

Great, oh, there was oh, the other thing is intra news. Now, where do we put at the top?

A

I know uh infra had a couple outages last week intra had a couple of footages.

A

uh Make sure so we got to make sure that we uh status.jenkins.io needs more visibility.

A

Don't know how.

B

Yeah many thanks to tim and damian for quickly working through uh and all the infrastructure team for quickly working through the release problem that we had uh this morning. um So if we've released the weekly successfully, oh nice.

A

I didn't even know about that one um and then uh package.

A

uh Yeah, so I gotta- I gotta actually say this out loud, not just write it down, uh update so updates jenkins. I o had an outage this week. uh As far as I understand it has something to do with the signing certificate, not free gps, but the actual internal signing that got tracked down and fixed up but they're. You know it did happen and then packages objects. I own mirrors, oh duncan's, I o they got moved into our normal infrastructure um with the actual the mirroring software that actually does mirror properly.

A

um There's a few issues to do with https there. They got fixed relative quickly, but you know there was an outage and it didn't get documented so yeah. It was a good good week for everyone, but we do. We did find a lot of people just didn't know to look at status. It's not something that breaks very often, so we got to figure out a way to make that more clear.

B

Yeah, I see so so the so the status page was up to date, but people weren't able to find it or didn't know that it existed yeah pretty much.

A

Yeah, so we have the status page that keeps track of all the different stuff. The info exists and there was an uh an issue with packages. You can see yeah we had the various things, but no one knows it exists this page because we don't have very many outages. No one knows to check this when there is issues probably updates is in here as well.

A

B

So I don't know if we're linking to this anywhere but yeah.

A

That was one thing someone said is like. We should probably put a link to it on jenkins iio. You know, I don't really know what else we do any ideas for anyone. I mean there's only two of us on this call, but if anyone's reading watching this later uh any up any suggestions on how to improve that communication pipeline would be pretty good.

B

Yeah we could have like a dedicated twitter account for status updates, but that might be overkill.

A

Yeah I mean I.

B

Believe we don't need.

A

It too often yeah, I mean the number of actual issues is they're, usually generally pretty low just needs a restart once in a while. This last week has been a little rough, but then it goes past there. It's like 225 days ago.

A

You know 30 days ago, and most of them are just issues with the the main ci service which just gets so much traffic right, oh and artifactory, that we don't. We don't maintain right, yeah, okay, well, yeah! I don't. I don't have anything else to do. No! Thank you! So much! Okay, I'm gonna end the recording. Now I think somehow.

A

And no, I don't want to do this, I'm just going to end the meeting so bye.