mirrors/seaweedfs
1
0
Fork 0
mirror of https://github.com/seaweedfs/seaweedfs.git synced 2026-09-27 10:24:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

master: check the reported volume digest and ask for the list on a mismatch

Browse Source 
Compared after everything the heartbeat carried has been applied, so agreement
means the master is current rather than that nothing changed.

Servers reporting no digest are untouched, and a mismatch on a heartbeat that
already carried the full list is reported rather than answered: there is
nothing further to ask for, so asking again would loop. Nodes reporting one
volume id twice are skipped for the same reason.
This commit is contained in:
Chris Lu
2026-08-07 12:53:27 -07:00
parent 342bbc64bc
commit 536a334999
1 changed files with 46 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
weed/server/master_grpc_server.go
+46
View File
@@ -275,9 +275,55 @@ func (ms *MasterServer) SendHeartbeat(stream master_pb.Seaweed_SendHeartbeatServ
if len(message.NewVids) > 0 || len(message.DeletedVids) > 0 || len(message.NewEcVids) > 0 || len(message.DeletedEcVids) > 0 {
ms.broadcastToClients(&master_pb.KeepConnectedResponse{VolumeLocation: message})
}
// Checked after everything the heartbeat carried has been applied, so a
// match means the master is current, not that nothing changed.
if resend := ms.checkVolumeDigest(heartbeat, dn); resend {
if err := stream.Send(&master_pb.HeartbeatResponse{ResendFullVolumeList: true}); err != nil {
glog.Warningf("SendHeartbeat.Send resend request to %s:%d %v", dn.Ip, dn.Port, err)
return err
}
}
}
}
// checkVolumeDigest compares the digest a volume server reported against the
// master's own, and reports whether the master needs the full volume list to
// recover. Servers that report no digest are left alone: they still send the
// whole list every time.
func (ms *MasterServer) checkVolumeDigest(heartbeat *master_pb.Heartbeat, dn *topology.DataNode) bool {
if heartbeat.VolumeDigest == nil {
return false
}
// One volume id mounted on two disks is reported twice but stored once, so
// the digests cannot agree however often the list is resent. Asking would
// loop forever.
if dn.HasDuplicateVolumeIds() {
stats.MasterReceivedHeartbeatCounter.WithLabelValues("volumeDigestNotComparable").Inc()
return false
}
reported := heartbeat.GetVolumeDigest()
held := dn.VolumeDigest()
if held == reported {
stats.MasterReceivedHeartbeatCounter.WithLabelValues("volumeDigestMatch").Inc()
return false
}
stats.MasterReceivedHeartbeatCounter.WithLabelValues("volumeDigestMismatch").Inc()
// A heartbeat that already carried the full list has nothing more to give;
// asking again would just repeat. Say so instead, because at this point the
// two ends genuinely disagree about what the server holds.
if len(heartbeat.Volumes) > 0 || heartbeat.HasNoVolumes {
glog.Warningf("volume server %s reported digest %d after a full volume list, master holds %d",
dn.Url(), reported, held)
return false
}
glog.V(0).Infof("volume server %s reported digest %d, master holds %d: requesting the full volume list",
dn.Url(), reported, held)
return true
}
// KeepConnected keep a stream gRPC call to the master. Used by clients to know the master is up.
// And clients gets the up-to-date list of volume locations
func (ms *MasterServer) KeepConnected(stream master_pb.Seaweed_KeepConnectedServer) error {