[
https://issues.apache.org/jira/browse/HDDS-4336?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Stephen O'Donnell updated HDDS-4336:
------------------------------------
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:
{code}
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)
{code}
The assertion here is what is failing, as it does not allow for the sequenceID
to be changed on a CLOSED container:
{code}
public void updateSequenceId(long sequenceID) {
assert (isOpen() || state == HddsProtos.LifeCycleState.QUASI_CLOSED);
sequenceId = max(sequenceID, sequenceId);
}
{code}
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.
was:
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 null, and then container
reports for the replica fail with a stack trace like:
{code}
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)
{code}
The assertion here is what is failing, as it does not allow for the sequenceID
to be changed on a CLOSED container:
{code}
public void updateSequenceId(long sequenceID) {
assert (isOpen() || state == HddsProtos.LifeCycleState.QUASI_CLOSED);
sequenceId = max(sequenceID, sequenceId);
}
{code}
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.
> ContainerInfo does not persist BCSID leading to failed replicas reports
> -----------------------------------------------------------------------
>
> Key: HDDS-4336
> URL: https://issues.apache.org/jira/browse/HDDS-4336
> Project: Hadoop Distributed Data Store
> Issue Type: Bug
> Components: SCM
> Affects Versions: 1.1.0
> Reporter: Stephen O'Donnell
> Assignee: Stephen O'Donnell
> Priority: Major
>
> 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:
> {code}
> 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)
> {code}
> The assertion here is what is failing, as it does not allow for the
> sequenceID to be changed on a CLOSED container:
> {code}
> public void updateSequenceId(long sequenceID) {
> assert (isOpen() || state == HddsProtos.LifeCycleState.QUASI_CLOSED);
> sequenceId = max(sequenceID, sequenceId);
> }
> {code}
> 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.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
