[
https://issues.apache.org/jira/browse/HDFS-10817?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15450558#comment-15450558
]
Zhe Zhang commented on HDFS-10817:
----------------------------------
Thanks [~xkrogen] for the work! Patch looks pretty good. A few minor issues:
# Some lines are longer than 80 chars. Also I think multi-line comments should
be:
{code}
/**
* blah blah
*/
{code}
# Can we unify the logic for read and write locking? I.e. {{lastUnlock}} and
{{needReport}}
# {{readLockHeldTimeStamp.remove();}} doesn't look necessary?
# It's a little flaky to only assert the log message contains the method name.
Can we do more precise assertion here?
# Retrospectively, we should probably consolidate both write and read lock
time-keeping in the {{FSNamesystemLock}} class itself. Doesn't make sense for
both {{writeLock}} and {{writeLockInterruptibly}} to do the check and count
time. We should file a follow-on JIRA for that.
# Another nice follow-on is to start a number of threads, some with sleep
(longer than the threshold) and some don't. Then just verify the output has the
ones with sleep.
> Add Logging for Long-held NN Read Locks
> ---------------------------------------
>
> Key: HDFS-10817
> URL: https://issues.apache.org/jira/browse/HDFS-10817
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: logging, namenode
> Reporter: Erik Krogen
> Assignee: Erik Krogen
> Attachments: HDFS-10817.000.patch, HDFS-10817.001.patch
>
>
> Right now the Namenode will log when a write lock is held for a long time to
> help tracks methods which are causing expensive delays. Let's do the same for
> read locks since these operations may also be expensive/long and cause
> delays.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
