[ https://issues.apache.org/jira/browse/HDFS-4477?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Daryn Sharp updated HDFS-4477: ------------------------------ Attachment: HDFS-4477.patch Untested, proposed patch, for feedback. I think the solution is simple, but with the complexity of locking and HA interactions, I need feedback: * Add a hook method for subclasses of ADTSM to intercept token expiration * Modify NN's DTSM to notify FSN of the expiration * Add method to FSN that generates a cancel token edit op The new FSN method is unsynchronized because it will be called from the ADTSM's background thread. The edit log op builders are thread-local, and the logEdit is synch'ed, so I think it's ok and preferable to not lock the whole namespace. > Secondary namenode may retain old tokens > ---------------------------------------- > > Key: HDFS-4477 > URL: https://issues.apache.org/jira/browse/HDFS-4477 > Project: Hadoop HDFS > Issue Type: Bug > Components: security > Affects Versions: 0.23.7 > Reporter: Kihwal Lee > Assignee: Daryn Sharp > Attachments: HDFS-4477.patch > > > Upon inspection of a fsimage created by a secondary namenode, we've > discovered it contains very old tokens. These are probably the ones that were > not explicitly canceled. It may be related to the optimization done to avoid > loading fsimage from scratch every time checkpointing. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira