Skip to content

Roll out: merge the four cluster PRs, set the 300s timeout, rebuild, restart #43

Closed

Description

@possiblyneal

Parent

Part of #31.

Branch: main

What to build

The fixes are in effect on the Operator's daemon, not just on main. Merge the four cluster PRs; set AGENTMEMORY_LLM_TIMEOUT_MS=300000 in ~/.agentmemory/.env (basis: worst measured cold 50k chunk at concurrency 2 is 152.8s, cold model load 9–16s, worst scheduler wait under 30s; this assumes the concurrency default of 2 set in #33 — do not raise SUMMARIZE_CHUNK_CONCURRENCY in .env without re-measuring); npm run build; systemctl --user restart agentmemory.service; confirm on the health endpoint that the running version is the merged build and that the summarize failure counter stops climbing on the previously poisoned Sessions.

Acceptance criteria

Blocked by

Activity

  1. possiblyneal commented on Sep 22, 2026

    @possiblyneal
    OwnerAuthor

    Rolled out: PRs #45–#48 and #50 merged (0019bef head), AGENTMEMORY_LLM_TIMEOUT_MS=300000 set with the reasoning_effort note corrected, dist rebuilt and agentmemory.service restarted (per-operation breakers visible on /health, 132 endpoints, healthy), poisoned Session 24d3db55 (500 Observations) summarized end to end in 467s as 4 chunks at concurrency 2 with no chunk failure or timeout (qualityScore 100), and the eleven triage rows moved to already-fixed in 7d177ee.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    ready-for-agentFully specified, ready for an AFK agent

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions