Andrew Scribner

fwiw, I've found the [jlumbroso/free-disk-space](https://github.com/jlumbroso/free-disk-space) action with default settings is working better, leaving a runner with ~45GB free after execution. An example of this is in [kfp's tests](https://github.com/canonical/kfp-operators/pull/428)

@rimolive can you remove @DnPlas from Charmed Kubeflow and replace her with myself? ty! to your questions, for Charmed Kubeflow: - Are you planning on having your distro ready in...

@Ayush9026 are you still working on this, or do we need additional help?

I think the root cause of this is #7552, where the 0th `matchLabel` is being used to "label" each `PodDefault` instead of the `PodDefault`'s name

This bug is also the cause of #6812, which identified how a the jwa fails to handle `PodDefault`s that have no `matchLabels` selector

This bug appears to also affect the tensorboard web app backend, which has a copy of the same `get_poddefaults()`

I don't understand typescript enough to be certain that the `label` is intended to be the name of the PodDefault, but given the comments in the spawner_ui_config.yaml I think that's...

I hit this bug today too, but I think the issue is slightly different (details in #7552). Rather than handle both `matchLabels` and `matchExpressions`, I believe the `label` returned here...

sorry coming to this late... > I still believe this will be simpler to reason about if ops reinitialised the charm before every hook (deferred and Juju events). That's what...

I bet this needs refactoring to be relevant, but that it will actually be useful in the next few weeks as our bundle tests come back online