leonarski_fandClaude Opus 4.8
9b1eeb1342
Build Packages / build:rpm (ubuntu2404_nocuda) (push) Successful in 10m3s
Build Packages / build:rpm (rocky8_nocuda) (push) Successful in 11m14s
Build Packages / build:rpm (rocky9_nocuda) (push) Successful in 11m34s
Build Packages / build:rpm (ubuntu2204_nocuda) (push) Successful in 11m32s
Build Packages / build:rpm (rocky8_sls9) (push) Successful in 11m42s
Build Packages / build:rpm (rocky9_sls9) (push) Successful in 12m30s
Build Packages / XDS test (durin plugin) (push) Successful in 6m49s
Build Packages / XDS test (JFJoch plugin) (push) Successful in 8m52s
Build Packages / build:rpm (rocky8) (push) Successful in 11m24s
Build Packages / build:rpm (ubuntu2204) (push) Successful in 10m35s
Build Packages / Create release (push) Skipped
Build Packages / Generate python client (push) Successful in 25s
Build Packages / Build documentation (push) Successful in 49s
Build Packages / build:rpm (rocky9) (push) Successful in 12m7s
Build Packages / build:rpm (ubuntu2404) (push) Successful in 11m0s
Build Packages / DIALS test (push) Successful in 13m11s
Build Packages / XDS test (neggia plugin) (push) Successful in 6m20s
Build Packages / Unit tests (push) Successful in 1h7m39s
A slow filesystem could stall the writer's receive pipeline, propagating TCP backpressure to the pusher. The pusher's 3s per-frame send deadline then treated that backpressure as a dead peer and force-closed the connection mid-run. The writer reconnected as a brand-new connection that was not part of the active session, so the rest of the run was silently dropped and the half-written HDF5 file was later finalized with holes. Replace the throughput-based send timeout with a peer-liveness timeout: - Add TCPFrameType::BUSY (wire version 2 -> 3). - TCPImagePuller runs a heartbeat thread that sends BUSY every 250ms on a thread independent of the stalled write path, so liveness keeps flowing during deep stalls. A send_mutex serializes ACK/pong/heartbeat writes. - TCPStreamPusher refreshes last_peer_activity on every inbound frame and only declares a connection dead after peer_liveness_timeout (15s) of complete silence, tolerating arbitrarily long backpressure while still catching a genuinely dead peer (and immediate EPIPE/ECONNRESET). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>