You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
fix: send each worker's actor status instead of joining it in the page
The workers bar coloured its segments from the loaded actor page, so searching,
sorting or paging the *actors* table recoloured the *workers* one — a pod whose
actor fell off the page read as `Not loaded`, a status no cluster reports. The
controller has both whole lists when it builds the response, so it carries the
status on the worker entry and the page joins nothing.
Also from review:
- Every actor is offered to the join, not only those the atespace scope kept. A
worker running an actor whose template lives elsewhere reported itself idle.
- A live actor on a shared pod wins over a parked one; the lowest id breaks ties.
- A failed read no longer renders as "No actors in this scope", and the legend
stays when a scope empties so the table beneath it does not jump.
- The per-actor ceiling is 80, which is the widest row that fits the card.
- An extension's PNG favicon is no longer served under an SVG type, and the
created link carries the marker so a second call retargets it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Nicholas Bucher <behappy54321@gmail.com>
0 commit comments