damccorm commented on code in PR #22225:
URL: https://github.com/apache/beam/pull/22225#discussion_r919070657


##########
sdks/go/pkg/beam/util/diagnostics/diagnostics.go:
##########
@@ -0,0 +1,68 @@
+// Licensed to the Apache Software Foundation (ASF) under one or more
+// contributor license agreements.  See the NOTICE file distributed with
+// this work for additional information regarding copyright ownership.
+// The ASF licenses this file to You under the Apache License, Version 2.0
+// (the "License"); you may not use this file except in compliance with
+// the License.  You may obtain a copy of the License at
+//
+//    http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package diagnostics
+
+import (
+       "bufio"
+       "context"
+       "os"
+
+       "runtime/debug"
+
+       "github.com/apache/beam/sdks/v2/go/pkg/beam/io/filesystem"
+       _ "github.com/apache/beam/sdks/v2/go/pkg/beam/io/filesystem/gcs"
+)
+
+func UploadHeapDump(ctx context.Context, dest string, additionalDataToWrite 
string) error {
+       heapDumpLoc := "heapDump"
+
+       f, err := os.Create(heapDumpLoc)
+       if err != nil {
+               return err
+       }
+       debug.WriteHeapDump(f.Fd())

Review Comment:
   I think you're right. I got my wires crossed as I moved this code around and 
thought it wasn't showing up because we never successfully allocate the array 
that causes the OOM. On closer inspection, even if we successfully allocate a 
large array in advance, it still won't show up correctly. 
   
   Just about the only path forward at this point would seem to be monitoring 
memory usage from a separate process and uploading a heap profile when we get 
close to full consumption. That's probably expensive though, and its not 100% 
clear to me we want to do that at the cost of performance. We at least will 
want to gate it behind an experiment.
   
   I'll think a little bit harder about this one



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to