Nick Mitchell

Howdy! Some terminals seem to be offering an ansi extension to support terminal links. ref: https://iterm2.com/documentation-escape-codes.html the logic i think is something along these lines: ```jsx const m = source.match(/\x1B\]8;;(.+)\u0007(.+)\x1B\]8;;\u0007/)...

## 🐛 Bug Module (check all that applies): * [ ] `torchx.spec` * [ ] `torchx.component` * [ ] `torchx.apps` * [ ] `torchx.runtime` * [x] `torchx.cli` * [ ]...

ray scheduler either appends a random string to my specified app_id, or ignores it entirely

4

## 🐛 Bug When I submit, e.g. ```shell torchx run -s ray -cfg app_id=myOwnUniqueString ... ``` it seems that the ray scheduler [always appends](https://github.com/pytorch/torchx/blob/main/torchx/schedulers/ray_scheduler.py#L231) a "unique" 64-bit suffix to whatever...

deployment -> pods uses incorrect label selector

1

kui uses `-l sel1 -l sel2` instead of `-l sel1,sel2` since it wants a conjunction of selectors

Update kubectl Logs tab to use xterm's search, rather than our home-grown

4

ours is nice, in that it filters to show only matching lines. but 1) sometimes i want to see context 2) we have to maintain lots of extra state on...

kubectl get --watch tables should eventually remove terminated resources

2

It is annoying for these to accumulate for long-term watches.

Logs tab should track initContainers

2

Right now, we ignore these. I'm not sure why, yet.

the dataset-operator logs show lots of this: > 1.7105266916204755e+09 ERROR controller.dataset Reconciler error {"reconciler group": "com.ie.ibm.hpsys", "reconciler kind": "Dataset", "name": "xxxx", "namespace": "yyyy", "error": "datasetsinternal.com.ie.ibm.hpsys \"xxxx\" is forbidden: unable to...

read-write Dataset gives `Socket not connected` errors after a `cp -r`

4

i am seeing issues when i try to mount a IBM Cloud COS bucket *read-write*. i see this error from pods that try to mount read-write Datasets: > Socket not...