From WhoKnows444Done 6 days agovia agent-mcp
2026-10-16 16:21Z, project Integration_v1, machine_1. Console agent-1790016811797-78sn0c (ticket e29f4bed) reports backlog_internal_comment returned nothing for 1800 s and was aborted (its last_text 16:05:03Z); HAIKU SONIC saw the same timeouts on two other seats at 14:39Z and 15:35Z. On my own console (agent-1790806630796-cd3fto) backlog_get 710a252e-a906-4309-bd24-eb14196bf418 has been running since 16:16Z (moved to background after 120 s, no result yet) while backlog_internal_comment on the same ticket answered in under a second at 16:18:54Z. So the backlog server is partly responsive: some calls hang, others on the same ticket return. 41 claude consoles live, renderer 2.06 GB, machine 74 percent free. No workaround applied; seats are told to retry and keep working from their hand-off files.
This will be available in the next version (part of it already shipped in 1.219.0): reading a single ticket used to download the whole backlog, which slowed our server down under load and made some backlog calls wait for minutes; that is fixed, and a call can no longer stay stuck: it now stops with a clear error after two minutes at most. If it still happens after updating, please reopen this ticket with as many details as possible (a screenshot or a video helps a lot).
Your browser sent this along so the team can reproduce what you saw. Nothing here was typed by you.