Add 2026.06 symlink#266
Conversation
casparvl
changed the title
|
bot: status |
|
This is the status of all the
|
3 similar comments
|
This is the status of all the
|
|
This is the status of all the
|
|
This is the status of all the
|
casparvl
changed the title
|
bot: build repo:eessi.io-2026.06-software instance:eessi-bot-mc-aws for:arch=x86_64/amd/zen2 |
|
New job on instance
|
|
New job on instance
|
Seems fine, up until the loading of the module. |
|
bot: build repo:eessi.io-2026.06-software instance:eessi-bot-mc-aws for:arch=x86_64/amd/zen2 |
|
New job on instance
|
…cripts into 2026.06-module
|
bot:cancel job:178715 |
|
bot: build repo:eessi.io-2026.06-software instance:eessi-bot-mc-aws for:arch=x86_64/amd/zen2 |
|
New job on instance
|
… UMASK. That doesn't make sense for some scripts that NEED to be executable, such as
|
bot: build repo:eessi.io-2026.06-software instance:eessi-bot-mc-aws for:arch=x86_64/amd/zen2 |
|
New job on instance
|
|
Don't want to wait for the node to boot and another one is free by now... bot:cancel job:178719 |
|
bot: build repo:eessi.io-2026.06-software instance:eessi-bot-mc-aws for:arch=x86_64/amd/zen2 |
|
New job on instance
|
|
Instance
|
|
Instance
|
|
Instance
|
|
bot: build repo:eessi.io-2026.06-software instance:eessi-bot-mc-aws for:arch=x86_64/amd/zen3 |
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
New job on instance
|
|
bot: build repo:eessi.io-2026.06-software instance:eessi-bot-riscv for:arch=riscv64/generic |
|
New job on instance
|
|
I guess this is what I get for pushing to have risc-v in the |
|
Ehhhh, from the logs: Maybe it's some of the custom logic that got inserted into the modulefile for RISC-V that is now causing trouble? |
|
Hmmm, That's effectively the same, maybe that's the duplication it's complaining about. |
|
I'm guessing it's because special things are in the build scripts for RISC-V, with the assumption that things are happening under |
|
Yeah, I have a few meetings, I'm hoping to spend the last hour of today to see if I can adapt it... I.e. do the special casing for EESSI/2025.06 + RISCv, but remove the special casing for EESSI/2026.06 |
|
Indeed, the riscv bot is using a different script: https://github.com/EESSI/dev.eessi.io-scripts/blob/main/bot/bot-build-dev.eessi.io.slurm. We may need some wrapper that determines what kind of build this is, and which script needs to be called. edit: actually, I had added something like that for compat layer builds already. Still need to push that to github, but the script on the bot account (in We could probably extend that and do something similar for 2026.06 repos. |
|
Oh, I thought it was just the customization in the EESSI modulefile, but I didn't realize it had build script that's actually different. Now that you mention it, I remember you told me this before when we were preparing for the 'site building on top of EESSI' webinar though... |
|
We can do a wrapper, but... maybe we should consider just launching a different bot instance for this? I.e. one that does NOT use the script from Or we should just figure out how to leverage the new site-building functionality to have the bot build in dev.eessi.io - I think during development of the webinar material we concluded this should be possible (and is likely preferable), but we didn't have time to implement it back then... |
5 participants
We should discuss in tomorrow's support meeting if we want to start exposing
2026.06.Also, before building, we need the bots to be configured to build for 2026.06 (https://github.com/EESSI/bot-configs/pull/97)
Edit: this was discussed in the support meeting today, there was consensus to not yet put this symlink in place, we only put the module in place in the
/version/2026.06specific directory. I think that will be created if we trigger builds on EESSI/software-layer#1546Edit2: my bad, we can (and should) do this PR. The symlink in
/cvmfs/software.eessi.io/initis made from the filesystem-layer side. This PR would create a symlink in/cvmfs/software.eessi.io/versions/2026.06/init/modules/EESSI, which is indeed what we need right now :)