fix: sync replica entries before ec.encode and volume.tier.move (#7798)

* fix: sync replica entries before ec.encode and volume.tier.move (#7797)

This addresses the data inconsistency risk in multi-replica volumes.

When ec.encode or volume.tier.move operates on a multi-replica volume:
1. Find the replica with the highest file count (the 'best' one)
2. Copy missing entries from other replicas INTO this best replica
3. Use this union replica for the destructive operation

This ensures no data is lost due to replica inconsistency before
EC encoding or tier moving.

Added:
- command_volume_replica_check.go: Core sync and select logic
- command_volume_replica_check_test.go: Test coverage

Modified:
- command_ec_encode.go: Call syncAndSelectBestReplica before encoding
- command_volume_tier_move.go: Call syncAndSelectBestReplica before moving

Fixes #7797

* test: add integration test for replicated volume sync during ec.encode

* test: improve retry logic for replicated volume integration test

* fix: resolve JWT issue in integration tests by using empty security.toml

* address review comments: add readNeedleMeta, parallelize status fetch, fix collection param, fix test issues

* test: use collection parameter consistently in replica sync test

* fix: convert weed binary path to absolute to work with changed working directory

* fix: remove skip behavior, keep tests failing on missing binary

* fix: always check recency for each needle, add divergent replica test
This commit is contained in:
Chris Lu
2025-12-16 23:16:07 -08:00
committed by GitHub
parent 9c4a2e1b1a
commit 347ed7cbfa
6 changed files with 959 additions and 15 deletions

View File

@@ -102,6 +102,9 @@ func (c *commandVolumeTierMove) Do(args []string, commandEnv *CommandEnv, writer
}
fmt.Printf("tier move volumes: %v\n", volumeIds)
// Collect volume ID to collection name mapping for the sync operation
volumeIdToCollection := collectVolumeIdToCollection(topologyInfo, volumeIds)
_, allLocations := collectVolumeReplicaLocations(topologyInfo)
allLocations = filterLocationsByDiskType(allLocations, toDiskType)
keepDataNodesSorted(allLocations, toDiskType)
@@ -143,7 +146,8 @@ func (c *commandVolumeTierMove) Do(args []string, commandEnv *CommandEnv, writer
}
for _, vid := range volumeIds {
if err = c.doVolumeTierMove(commandEnv, writer, vid, toDiskType, allLocations); err != nil {
collection := volumeIdToCollection[vid]
if err = c.doVolumeTierMove(commandEnv, writer, vid, collection, toDiskType, allLocations); err != nil {
fmt.Printf("tier move volume %d: %v\n", vid, err)
}
allLocations = rotateDataNodes(allLocations)
@@ -192,7 +196,7 @@ func isOneOf(server string, locations []wdclient.Location) bool {
return false
}
func (c *commandVolumeTierMove) doVolumeTierMove(commandEnv *CommandEnv, writer io.Writer, vid needle.VolumeId, toDiskType types.DiskType, allLocations []location) (err error) {
func (c *commandVolumeTierMove) doVolumeTierMove(commandEnv *CommandEnv, writer io.Writer, vid needle.VolumeId, collection string, toDiskType types.DiskType, allLocations []location) (err error) {
// find volume location
locations, found := commandEnv.MasterClient.GetLocationsClone(uint32(vid))
if !found {
@@ -208,12 +212,18 @@ func (c *commandVolumeTierMove) doVolumeTierMove(commandEnv *CommandEnv, writer
if isOneOf(dst.dataNode.Id, locations) {
continue
}
var sourceVolumeServer pb.ServerAddress
for _, loc := range locations {
if loc.Url != dst.dataNode.Id {
sourceVolumeServer = loc.ServerAddress()
}
// Sync replicas and select the best one (with highest file count) for multi-replica volumes
// This addresses data inconsistency risk in multi-replica volumes (issue #7797)
// by syncing missing entries between replicas before moving
sourceLoc, selectErr := syncAndSelectBestReplica(
commandEnv.option.GrpcDialOption, vid, collection, locations, dst.dataNode.Id, writer)
if selectErr != nil {
fmt.Fprintf(writer, "failed to sync and select source replica for volume %d: %v\n", vid, selectErr)
continue
}
sourceVolumeServer := sourceLoc.ServerAddress()
if sourceVolumeServer == "" {
continue
}