Skip to content

Report CouchDB LiveSync startup failures and timestamp replication statistics - #232

Merged
vrtmrz merged 11 commits into
mainfrom
fix/continuous-network-recovery
Nov 20, 2026
Merged

vrtmrz merged 11 commits into
mainfrom
fix/continuous-network-recovery

Conversation

@vrtmrz

@vrtmrz vrtmrz commented Nov 19, 2026

Copy link
Copy Markdown
Owner

This change gives hosts the reason a CouchDB LiveSync start failed and records when replication statistics were last published.

Behaviour

  • A host can observe one outcome for preparation and the initial handoff to continuous transfer. Confirmed network failures in readiness, connection, or the initial pull carry the observed host state and time.
  • An explicit stop is reported when the service receives the request, including when startup has yet to produce an active transfer.
  • Concurrent synchronisation parameter reads share the remote operation. A failed read retains its cause, and an older refresh cannot discard a newer cached result. Concurrent fetch() callers can receive distinct Promise objects for the shared read.
  • The statistics publisher records an epoch-millisecond timestamp. Selecting a new Replicator resets it to null.

Verification

  • All 4,221 unit tests passed, along with type, dependency-boundary, release-tool, and packed-package checks.
  • Self-hosted LiveSync was built with the packed candidate. A Linux Obsidian and CouchDB run recovered a waiting file after an offline restart and received a subsequent edit over continuous synchronisation.

Refs vrtmrz/obsidian-livesync#1224.

@vrtmrz
vrtmrz merged commit e429a43 into main Nov 20, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant