search

quarkusio / quarkus-images

Set of container images delivered for Quarkus
Apache License 2.0
110 stars 75 forks source link

Mandrel 21.3.3.0-Final and 22.2.0.0-Final July 2022 CPU #202

Closed Karm closed 1 year ago

Karm commented 1 year ago

@zakkak @cescoffier https://github.com/graalvm/mandrel/releases/tag/mandrel-22.2.0.0-Final might have been in DRAFT state the second this CI was started, so if the download fails, restart the CI, please...

Karm commented 1 year ago

LGTM. Thanks @Karm!

I think we should remove 22.0 mandrel modules at this point, in case you want to squeeze it in the same commit.

How about we do a cleanup with the LTS? Like "when LTS comes, all previous in that major are dropped".

zakkak commented 1 year ago

How about we do a cleanup with the LTS? Like "when LTS comes, all previous in that major are dropped".

I recall having some issues with keeping that many images up to date, but I don't recall what the issue was. @cescoffier might have a better memory :)

cescoffier commented 1 year ago

The deployment will fail, unfortunately. The whole deployment process is broken since multi-archs. Look at the multi-archs branch, which proposes a completely new approach to building the images.

Karm commented 1 year ago

The deployment will fail, unfortunately. The whole deployment process is broken since multi-archs. Look at the multi-archs branch, which proposes a completely new approach to building the images.

I've seen the multi-archs branch. Not sure I understand how it's meant to work though.

@cescoffier Could you help me to understand what would be the best now, please?

Having updated images up and running on quay is definitely a higher priority for us than having multiarch.

Should I remove the Mandrel multiarch from the current branch, do the update and iterate async later in the multi-arch branch?

galderz commented 1 year ago

... The whole deployment process is broken since multi-archs...

🙈

galderz commented 1 year ago

@cescoffier if my gift to you (https://github.com/quarkusio/quarkus-images/commit/21e25a3c3ae3e3f4a4bd02b1e76f6051c31729c0) is to problematic, why not revert it? As I already indicated in https://github.com/quarkusio/quarkus-images/pull/196, https://github.com/quarkusio/quarkus-images/commit/21e25a3c3ae3e3f4a4bd02b1e76f6051c31729c0 is only half of the job. You need the (hacked) code in the PR to be able to push things. So, I'd be very glad for my work to be reverted and unlock @Karm's and @gastaldi's PRs.

galderz commented 1 year ago

https://github.com/quarkusio/quarkus-images/pull/203 needs to be integrated before this PR can. Also, once integrated this PR might need to be reworked.

Karm commented 1 year ago

@galderz @zakkak I rebased, removed multiarch naming and also dropped old images.

Karm commented 1 year ago

~O.K., that removal part was not a good move "Unable to find cekit module for version 21.2.0.2-Final-java11 - exiting"~

fixed...

Karm commented 1 year ago

Can we have the rebuild triggered now?

https://quay.io/repository/quarkus/ubi-quarkus-mandrel?tab=tags

galderz commented 1 year ago

@Karm Probably but I would maybe prefer to rebuild both Mandrel and GraalVM images at the same time so that they're available (roughly) at the same time. GraalVM PRs look good but still need merging (I can't do that).