[HDFS-5080] BootstrapStandby not working with QJM when the existing NN is active - ASF JIRA

Voters

Watch issue

Watchers

Create sub-task

Link

Clone

Update Comment Author

Replace String in Comment

Update Comment Visibility

Delete Comments

XML

Word

Printable

JSON

Details

Type: Bug
Status: Closed
Priority: Major
Resolution: Fixed
Affects Version/s: 3.0.0-alpha1
Fix Version/s: 2.1.1-beta
Component/s: ha, qjm
Labels:
None

Hadoop Flags:

Reviewed

Description

Currently when QJM is used, running BootstrapStandby while the existing NN is active can get the following exception:

FATAL ha.BootstrapStandby: Unable to read transaction ids 6175397-6175405 from the configured shared edits storage. Please copy these logs into the shared edits storage or call saveNamespace on the active node.
Error: Gap in transactions. Expected to be able to read up until at least txid 6175405 but unable to find any edit logs containing txid 6175405
java.io.IOException: Gap in transactions. Expected to be able to read up until at least txid 6175405 but unable to find any edit logs containing txid 6175405
	at org.apache.hadoop.hdfs.server.namenode.FSEditLog.checkForGaps(FSEditLog.java:1300)
	at org.apache.hadoop.hdfs.server.namenode.FSEditLog.selectInputStreams(FSEditLog.java:1258)
	at org.apache.hadoop.hdfs.server.namenode.ha.BootstrapStandby.checkLogsAvailableForRead(BootstrapStandby.java:229)

Looks like the cause of the exception is that, when the active NN is queries by BootstrapStandby about the last written transaction ID, the in-progress edit log segment is included. However, when journal nodes are asked about the last written transaction ID, in-progress edit log is excluded. This causes BootstrapStandby#checkLogsAvailableForRead to complain gaps.

To fix this, we can either let journal nodes take into account the in-progress editlog, or let active NN exclude the in-progress edit log segment.