[hero_planner_admin] dashboard broken after oschema cutover (Methods —, disconnected) #21

Closed
opened 2026-06-14 15:00:50 +00:00 by sameh-farouk · 1 comment
Member

After cutting the admin to the oschema build, the dashboard rendered but showed Methods — and a disconnected indicator (screenshot), with console 404/502s for the spec + health.

Three root causes (planner-owned):

  1. PEER was set to the crate name hero_planner_server; the rpc.sock lives under the service name hero_planner, so PeerTarget pointed at a nonexistent socket -> domain discovery + the openrpc proxy 502d.
  2. the admin served /health but the shared connection JS polls /health.json -> 404.
  3. rpc_url was the stale /rpc-peer -> 404.

Tracked as already-fixed locally; resolution comment to follow at merge.

screenshot

After cutting the admin to the oschema build, the dashboard rendered but showed `Methods —` and a disconnected indicator (screenshot), with console 404/502s for the spec + health. **Three root causes (planner-owned):** 1. `PEER` was set to the crate name `hero_planner_server`; the `rpc.sock` lives under the **service** name `hero_planner`, so `PeerTarget` pointed at a nonexistent socket -> domain discovery + the openrpc proxy 502d. 2. the admin served `/health` but the shared connection JS polls `/health.json` -> 404. 3. `rpc_url` was the stale `/rpc-peer` -> 404. Tracked as already-fixed locally; resolution comment to follow at merge. ![screenshot](https://forge.ourworld.tf/attachments/3975a587-92da-41a5-8be3-3128523b0088)
Author
Member

Fixed on development in 0c051c4.

Three planner-owned root causes resolved:

  1. PEER set to the service/socket name hero_planner (was the crate name hero_planner_server) -> domain discovery + openrpc proxy restored.
  2. admin now serves /health.json (the path the shared connection JS polls).
  3. rpc_url moved off the dead /rpc-peer.

Admin dashboard restored: discovery works, 97-method spec loads, 0 console errors. (Follow-on rpc_url domain-scoping landed in #19 / 7456769.)

Fixed on `development` in `0c051c4`. Three planner-owned root causes resolved: 1. `PEER` set to the service/socket name `hero_planner` (was the crate name `hero_planner_server`) -> domain discovery + openrpc proxy restored. 2. admin now serves `/health.json` (the path the shared connection JS polls). 3. `rpc_url` moved off the dead `/rpc-peer`. Admin dashboard restored: discovery works, 97-method spec loads, 0 console errors. (Follow-on `rpc_url` domain-scoping landed in #19 / `7456769`.)
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
lhumina_code/hero_planner#21
No description provided.