fix: allow cancel of decom only when its in progress

cmd/erasure-server-pool-decom.go

@@ -245,6 +245,7 @@ func (p *poolMeta) QueueBuckets(idx int, buckets []decomBucketInfo) {
 var (
 	errDecommissionAlreadyRunning = errors.New("decommission is already in progress")
 	errDecommissionComplete       = errors.New("decommission is complete, please remove the servers from command-line")
+	errDecommissionNotStarted     = errors.New("decommission is not in progress")
 )
 
 func (p *poolMeta) Decommission(idx int, pi poolSpaceInfo) error {
@@ -1183,7 +1184,11 @@ func (z *erasureServerPools) Status(ctx context.Context, idx int) (PoolStatus, e
 	poolInfo := z.poolMeta.Pools[idx]
 	if poolInfo.Decommission != nil {
 		poolInfo.Decommission.TotalSize = pi.Total
-		poolInfo.Decommission.CurrentSize = poolInfo.Decommission.StartSize + poolInfo.Decommission.BytesDone
+		if poolInfo.Decommission.Failed || poolInfo.Decommission.Canceled {
+			poolInfo.Decommission.CurrentSize = pi.Free
+		} else {
+			poolInfo.Decommission.CurrentSize = poolInfo.Decommission.StartSize + poolInfo.Decommission.BytesDone
+		}
 	} else {
 		poolInfo.Decommission = &PoolDecommissionInfo{
 			TotalSize:   pi.Total,
@@ -1219,15 +1224,21 @@ func (z *erasureServerPools) DecommissionCancel(ctx context.Context, idx int) (e
 	z.poolMetaMutex.Lock()
 	defer z.poolMetaMutex.Unlock()
 
+	fn := z.decommissionCancelers[idx]
+	if fn == nil {
+		// canceling a decommission before it started return an error.
+		return errDecommissionNotStarted
+	}
+
+	defer fn() // cancel any active thread.
+
 	if z.poolMeta.DecommissionCancel(idx) {
-		if fn := z.decommissionCancelers[idx]; fn != nil {
-			defer fn() // cancel any active thread.
-		}
 		if err = z.poolMeta.save(ctx, z.serverPools); err != nil {
 			return err
 		}
 		globalNotificationSys.ReloadPoolMeta(ctx)
 	}
+
 	return nil
 }
 
@@ -1245,8 +1256,9 @@ func (z *erasureServerPools) DecommissionFailed(ctx context.Context, idx int) (e
 
 	if z.poolMeta.DecommissionFailed(idx) {
 		if fn := z.decommissionCancelers[idx]; fn != nil {
-			defer fn() // cancel any active thread.
-		}
+			defer fn()
+		} // cancel any active thread.
+
 		if err = z.poolMeta.save(ctx, z.serverPools); err != nil {
 			return err
 		}
@@ -1269,8 +1281,9 @@ func (z *erasureServerPools) CompleteDecommission(ctx context.Context, idx int)
 
 	if z.poolMeta.DecommissionComplete(idx) {
 		if fn := z.decommissionCancelers[idx]; fn != nil {
-			defer fn() // cancel any active thread.
-		}
+			defer fn()
+		} // cancel any active thread.
+
 		if err = z.poolMeta.save(ctx, z.serverPools); err != nil {
 			return err
 		}

docs/bucket/replication/setup_2site_existing_replication.sh

@@ -84,7 +84,7 @@ remote_arn=$(./mc replicate ls sitea/bucket --json | jq -r .rule.Destination.Buc
 sleep 1
 
 ./mc replicate resync start sitea/bucket/ --remote-bucket "${remote_arn}"
-sleep 20s ## sleep for 20s idea is that we give 200ms per object.
+sleep 30s ## sleep for 30s idea is that we give 300ms per object.
 
 count=$(./mc replicate resync status sitea/bucket --remote-bucket "${remote_arn}" --json | jq .resyncInfo.target[].replicationCount)
 
@@ -99,7 +99,7 @@ if [ $ret -ne 0 ]; then
 fi
 
 if [ $count -ne 12 ]; then
-	echo "resync not complete after 10s unexpected failure"
+	echo "resync not complete after 30s - unexpected failure"
 	./mc diff sitea/bucket siteb/bucket
 	exit 1
 fi

docs/distributed/decom-compressed-sse-s3.sh

@@ -33,7 +33,7 @@ sleep 2
 ./mc admin policy create myminio/ lake ./docs/distributed/rw.json
 
 ./mc admin policy attach myminio/ rw --user=minio123
-./mc admin policy attach myminio/ lake,rw --user=minio12345
+./mc admin policy attach myminio/ lake --user=minio12345
 
 ./mc mb -l myminio/versioned
 

docs/distributed/decom-encrypted-sse-s3.sh

@@ -28,7 +28,7 @@ sleep 2
 ./mc admin policy create myminio/ lake ./docs/distributed/rw.json
 
 ./mc admin policy attach myminio/ rw --user=minio123
-./mc admin policy attach myminio/ lake,rw --user=minio12345
+./mc admin policy attach myminio/ lake --user=minio12345
 
 ./mc mb -l myminio/versioned
 

docs/distributed/decom-encrypted.sh

@@ -30,7 +30,7 @@ export MC_HOST_myminio="http://minioadmin:minioadmin@localhost:9000/"
 ./mc admin policy create myminio/ lake ./docs/distributed/rw.json
 
 ./mc admin policy attach myminio/ rw --user=minio123
-./mc admin policy attach myminio/ lake,rw --user=minio12345
+./mc admin policy attach myminio/ lake --user=minio12345
 
 ./mc mb -l myminio/versioned
 

docs/distributed/decom.sh

@@ -15,7 +15,7 @@ fi
 
 export CI=true
 
-(minio server /tmp/xl/{1...10}/disk{0...1} 2>&1 >/dev/null) &
+(minio server /tmp/xl/{1...10}/disk{0...1} 2>&1 >/tmp/decom.log) &
 pid=$!
 
 sleep 2
@@ -29,7 +29,7 @@ export MC_HOST_myminio="http://minioadmin:minioadmin@localhost:9000/"
 ./mc admin policy create myminio/ lake ./docs/distributed/rw.json
 
 ./mc admin policy attach myminio/ rw --user=minio123
-./mc admin policy attach myminio/ lake,rw --user=minio12345
+./mc admin policy attach myminio/ lake --user=minio12345
 
 ./mc mb -l myminio/versioned