Agent Of Empires Daily: Sessions That Won't Let Go
Today's activity centers on session lifecycle bugs — stale state, runaway pollers, and stale mounts that all boil down to cleanup logic not keeping pace with real-world edge cases. Six PRs landed fixes across the TUI, session polling, and sandbox containers.
Duration: PT2M37S
Episode overview
This episode is a short developer briefing from Agent Of Empires Daily.
It explains recent repository work in plain language.
- Show: Agent Of Empires Daily
- Published: 2026-09-06T20:00:26Z
- Audio duration: PT2M37S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Good morning. It's September 6th, and this is Agent Of Empires Daily.
The clear pattern today: session state is getting stuck in limbo, and multiple PRs are chasing down why.
Start with PR 3786. A session archived mid-permission-prompt keeps showing as "Waiting" forever — no pane, no process, no way to clear it. Author BTForIT traces this to two earlier fixes stepping on each other: one tears down the session's tmux on archive, the other stops the poller from touching archived rows.…
Related, PR 3787 tackles the session-id poller itself. The thread ceiling was hard-coded at fifty, and once a daemon exceeded that, it re-probed every over-budget session on every tick — logging failures at a rate of roughly two and a half lines per second. The fix makes that budget configurable and adds repair…
On the sandbox side, PR 3785 from clefru fixes a similar class of problem: containers using status hooks keep a bind mount pointing at a temp directory that doesn't survive a host reboot. Restarting the container then fails outright. The fix restores that mount source before restart, using the existing hardened helper.
Two smaller but related fixes: PR 3782 keeps the TUI's storage mirror…