daily-held: reap stranded old-socket tmux server (~55 GiB)
kind: infra#daily-held#needs-human
track:Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.
No comments yet.
Overview / Motivation
Auto-filed by the 2026-07-18 /daily problem sweep (route 3 — destructive / human-gated). Sources: chunk-0 Problem 10, chunk-5 Problem 16 (logs/daily/mining-2026-07-18/).
Goal
Reap the stranded pre-cutover shared-socket tmux server (~55 GiB RSS, left behind when mygoat-session.service moved to its dedicated tmux -L mygoat socket on 2026-07-18), and close out the one dangling thread from the 01:00 UTC fleet-kill: the killed interactive session's open Leg-3 delta-CI verification question (#825/#1345 work — "the report doesn't mention Leg 3 ... Verifying" at 00:58, never answered).
Held item
- What happened: the mygoat-session.service 12h recycle on the #1466 shared tmux socket killed every tmux session on the VM at 01:00:48 UTC (3 live Claude sessions + Thomas's interactive tmux). Root cause was fixed the same day (dedicated
-L mygoatsocket); the dropped experiment ask was re-captured as #1489. - Why held (carve-out): destructive / irreversible — killing a live (if orphaned) tmux server process; risk of taking a still-referenced session with it. Needs a human eyeball on
tmux -L <old-socket> lsbefore the kill. - Suggested action: verify nothing live is attached to the old server,
killit (frees ~55 GiB RSS), and confirm whether the Leg-3 verification was completed in a successor session or still needs a pass.
Provenance
- source: /daily 2026-07-18 problem sweep (route 3)
- verified-at-filing: project memory
project_mygoat_tmux_socket_isolation.md(2026-07-18) records the stranded-L oldserver pending cleanup; the fix (socket isolation) is live per the same memory + chunk-0 P10 evidence (2026-07-19)