hesela.devdev_loss_tmo (transport device loss)
In Linux's Fibre Channel transport, dev_loss_tmo is the interval in seconds for retaining a missing remote target before device-removal processing.
Linux v6.12 separates device-loss work from fail-fast work; target removal affects the operating system's devices, not erasure of the stored data. It does not establish whether an earlier write executed. [1]
The reviewed multipath manual applies this setting to FC and SAS, not every transport. With positive fast_io_fail_tmo and finite no_path_retry N, multipathd keeps dev_loss_tmo larger than N times polling_interval. Premature removal while I/O remains queued can deadlock. Inspect the effective settings together. [2]
These FC path-loss timers do not trigger if the remote port is never reported lost, even when SCSI commands hang. The manual discusses eh_deadline for that different condition and warns that its HBA reset affects all target paths on the adapter. [2]
Relations
- related
- fast_io_fail_tmo (FC path failure), no_path_retry, Multipath I/O, I/O timeout, Fibre Channel, Serial Attached SCSI (SAS)
Citations
- Linux v6.12: FC transport loss, fail-fast and target-removal work.
- multipath-tools: multipath.conf(5), no_path_retry and FC timeout options; pinned commit d53932bdbee02f440518ea384e63d7c4d3f66685.
Further reading
Sources & caveats
- The FC definition follows Linux v6.12. FC/SAS applicability and queue-budget coordination follow the pinned multipath-tools manual. Non-erasure and uncertain-write conclusions are editorial interpretation, not measured outcomes.