AI-Hypercomputer / AI-Hypercomputer/xpk

Error applying Kueue CRDs on MacOS

Open
#9 3 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
193
Forks
94
Avg merge
24m
Merged PRs (30d)
1

Description

(I will update this bug with more information to help reproduce this issue later)

**Environment:** MacOS

**Command:**

`python xpk.py cluster create --cluster xpk-test --tpu-type=v4-16 --num-slices 1 --cluster-cpu-machine-type=n1-standard-8`

Everything works as expected until we start applying the Kueue CRDs [here](https://github.com/google/xpk/blob/main/xpk.py#L983).

```
[XPK] Starting xpk
...
[XPK] Task: `Set Kueue On Cluster` terminated with code `0`
[XPK] Enable Kueue CRDs
[XPK] Try 1: Applying Kueue CRDs
[XPK] Task: `Applying Kueue CRDs` is implemented by `kubectl apply -f 3`, streaming output live.
[XPK] Waiting for `Applying Kueue CRDs`, for 0 seconds
[XPK] Waiting for `Applying Kueue CRDs`, for 1 seconds
error: the path "3" does not exist
```

As you can see, the temporary file name generated (`"3"`) seems to be invalid/garbage.

When running this same command on Linux, it works, and we see the following log line:

```[XPK] Task: `Applying Kueue CRDs` is implemented by `kubectl apply -f /tmp/tmppwercgir, streaming output live.```

This shows a valid temporary filename in the expected location, `/tmp`.

Right now, my hunch is it is one of the following:

A) There is a discrepancy in the way the `tempfile` package creates/handles/ temporary files on MacOS vs Linux.
B) Perhaps this is a weird manifestation of a permissions issue.
C) Some discrepancy in Python version being used, or version of `tempfile` package being used.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.