From 946a22803777060954f94bcbc3d953be5240d02f Mon Sep 17 00:00:00 2001 From: Jeremy Dahlgren Date: Fri, 23 May 2025 12:43:02 -0400 Subject: [PATCH] Allow missing shard stats for restarted nodes for _snapshot/_status Adds a note explaining the change made in elasticsearch PR #128399 to reduce latency when getting stats for currently running snapshots. Relates ES-10982 --- specification/snapshot/status/SnapshotStatusRequest.ts | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/specification/snapshot/status/SnapshotStatusRequest.ts b/specification/snapshot/status/SnapshotStatusRequest.ts index 613025958b..b8e12e5c3d 100644 --- a/specification/snapshot/status/SnapshotStatusRequest.ts +++ b/specification/snapshot/status/SnapshotStatusRequest.ts @@ -30,6 +30,11 @@ import { Duration } from '@_types/Time' * * If you omit the `` request path parameter, the request retrieves information only for currently running snapshots. * This usage is preferred. + * Note that if a node has been restarted or has left the cluster since completing a shard snapshot the stats for that shard will be unavailable. + * Loading the stats from the repository is an expensive operation (see the WARNING below), so to minimize latency for returning stats for currently + * running snapshots the stats values will be zero for these shards even though the "stage" value will be "DONE". A "description" field will be set + * on these shard stats instances indicating why they are empty. Note that the total stats for the index will be less than expected due to the + * missing values from these shards. * If needed, you can specify `` and `` to retrieve information for specific snapshots, even if they're not currently running. * * WARNING: Using the API to return the status of any snapshots other than currently running snapshots can be expensive.