Files
clawstor/claw-store/src/cluster
osobhandClaude Sonnet 5 e5fcb8b3f4
Build with clawstor cache / Cargo build (clawstor-cached) (pull_request) Failing after 3s
Fix stdout/stderr ordering in ShutdownPrepCheck
Command::output() captures stdout and stderr as two separate
buffers. check() was concatenating stdout-then-stderr, which throws
away chronological order entirely -- every stderr line (e.g.
"Error: send_to_host not set" from `claw-store replicate` on a node
with no downstream replication target, which is normal and expected
on architect) landed at the very end of the report regardless of
when it actually printed, making a mid-script, already-handled
condition look like a failure that happened after "DRY RUN
COMPLETE".

Fix: invoke via `bash -c "script --dry-run 2>&1"` so stderr merges
into stdout inside the shell, before either stream reaches us --
true chronological order preserved, single buffer to read.

Verified on architect: the send_to_host message now appears exactly
where it happens, inside the "taking final snapshot + replicating"
step, with DRY RUN COMPLETE correctly last.

Co-Authored-By: Claude Sonnet 5 <[email protected]>
2026-07-31 19:45:29 -07:00
..
2026-07-14 09:26:47 -07:00
2026-07-12 04:36:34 -07:00
2026-07-14 01:03:46 -07:00
2026-07-14 00:47:19 -07:00