[HDDS-4336] ContainerInfo does not persist BCSID leading to failed replicas reports

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.1.0
• Fix Version/s: 1.1.0
• Component/s: SCM
• Labels: pull-request-available
• Target Version/s: 1.1.0
• Assignee: Stephen O'Donnell
• Reporter: Stephen O'Donnell
• Created: 12/Oct/20 13:24
• Updated: 23/Oct/20 04:45
• Resolved: 13/Oct/20 09:40

Description

If you create a container, and then close it, the BCSID is synced on the datanodes and then the value is updated in SCM via setting the "sequenceID" field on the containerInfo object for the container.

If you later restart just SCM, the sequenceID becomes zero, and then container reports for the replica fail with a stack trace like:

Exception in thread "EventQueue-ContainerReportForContainerReportHandler" java.lang.AssertionError
	at org.apache.hadoop.hdds.scm.container.ContainerInfo.updateSequenceId(ContainerInfo.java:176)
	at org.apache.hadoop.hdds.scm.container.AbstractContainerReportHandler.updateContainerStats(AbstractContainerReportHandler.java:108)
	at org.apache.hadoop.hdds.scm.container.AbstractContainerReportHandler.processContainerReplica(AbstractContainerReportHandler.java:83)
	at org.apache.hadoop.hdds.scm.container.ContainerReportHandler.processContainerReplicas(ContainerReportHandler.java:162)
	at org.apache.hadoop.hdds.scm.container.ContainerReportHandler.onMessage(ContainerReportHandler.java:130)
	at org.apache.hadoop.hdds.scm.container.ContainerReportHandler.onMessage(ContainerReportHandler.java:50)
	at org.apache.hadoop.hdds.server.events.SingleThreadExecutor.lambda$onMessage$1(SingleThreadExecutor.java:81)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
	at java.lang.Thread.run(Thread.java:748)

The assertion here is failing, as it does not allow for the sequenceID to be changed on a CLOSED container:

  public void updateSequenceId(long sequenceID) {
    assert (isOpen() || state == HddsProtos.LifeCycleState.QUASI_CLOSED);
    sequenceId = max(sequenceID, sequenceId);
  }

The issue seems to be caused by the serialisation and deserialisation of the containerInfo object to protobuf, as sequenceId never persisted or restored.

However, I am also confused about how this ever worked, as this is a pretty significant problem.

Issue Links

• duplicates: HDDS-4384 Container sequence ID is not loaded after SCM restart
• links to: GitHub Pull Request #1488