Skip to content

docs(arc-enterprise): catch-up re-pulls a node's own files; the manifest sweep waits for it - #89

Merged
xe-nvdk merged 1 commit into
mainfrom
docs/catch-up-own-origin-repull
Sep 29, 2026
Merged

xe-nvdk merged 1 commit into
mainfrom
docs/catch-up-own-origin-repull

Conversation

@xe-nvdk

@xe-nvdk xe-nvdk commented Sep 29, 2026

Copy link
Copy Markdown
Member

Summary

  • Clustering page, catch-up section: since 26.09.2 the walker checks the files this node originated instead of assuming it still holds them, so a node restored with an empty data disk under the same cluster.node_id pulls its own files back from a peer's replica as part of catch-up; the reconciler's orphan-manifest sweep waits for that convergence and reports a held sweep as manifest_sweep_held.
  • The walker-disabled callout says that the same switch turns off both the re-pull and the hold, so reconciliation should stay in dry run after such a restore.

Docs-only; pairs with Basekick-Labs/arc#961 (fixes Basekick-Labs/arc#959).

…est sweep waits for it

Arc #961 (#959): on per-node storage the catch-up walker no longer
assumes a node still holds the files it originated. A node restored with an
empty data disk pulls its own files back from a peer's replica as part of
catch-up, the query gate covers them, and the reconciler's orphan-manifest
sweep waits for convergence, reporting a held sweep as manifest_sweep_held.
With the walker disabled neither applies; say so in the existing callout.
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.

reconciliation: a node restored from an empty disk never re-pulls its own-origin files, and an act-mode run then proposes manifest deletes for them

1 participant