Bugfix release for multi-pod logs: no more dead streams when a service
selector matches pods from other workloads, and pod colours you can
actually tell apart.
Why
Pressing l on a service tails every pod its selector matches. The
container name was taken from the first pod and forced on all of them —
so when the selector also catches pods from other workloads (cronjobs
sharing a label, for example), every mismatched pod died with
container app is not valid for pod … and the merged view filled up
with errors instead of logs.
What
Logs
- Per-pod container fallback: if the requested container doesn't exist
on a given pod, that pod streams its own first container instead of
erroring out. Verified live against a namespace where one service
selects 11 pods across 4 different workloads — all 11 stream clean. - Pod colours are now assigned by position from a hue-separated
palette (purple → green → orange → blue → pink → lime → yellow →
red → cyan → magenta) instead of an FNV hash of the pod name.
Neighbouring replicas always land on distinct hues — no more
yellow-next-to-orange guesswork.
Install
go install github.com/LywwKkA-aD/k4s/cmd/k4s@latest
Full changelog: https://k4scli.io/changelog/