[
https://issues.apache.org/jira/browse/HDFS-13769?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16573844#comment-16573844
]
Wei-Chiu Chuang edited comment on HDFS-13769 at 8/8/18 8:46 PM:
----------------------------------------------------------------
Thanks for the new revision.
* IMO, this jira should convert to a HADOOP jira. Trash is not a HDFS-specific
feature and this could also be used in other file systems as well (S3A for
example)
* Instead of calling fs.listStatus(), would you please use
fs.listStatusIterator()?
** The former gets *everything* under a path, so you would see a bump in JVM
heap usage for a large dir.
* I am still not satisfied with FileSystem#contentSummary(). The closest I
could find is FileSystem#getQuotaUsage() which would return number of objects
in a directory. but quota is not enabled by default.
*Nits:
{code}
import static org.apache.hadoop.fs.CommonConfigurationKeysPublic.*;
{code}
TrashPolicyWithSafeDelete should not do wildcard import
* Nits2:
{code}
LOG.debug("DIR "+ path + " in trash is too large, try safe delete.");
{code}
This is not necessarily true, if skipCheckLimit is true.
was (Author: jojochuang):
Thanks for the new revision.
IMO, this jira should convert to a HADOOP jira. Trash is not a HDFS-specific
feature and this could also be used in other file systems as well (S3A for
example)
Instead of calling fs.listStatus(), would you please use
fs.listStatusIterator()?
The former gets *everything* under a path, so you would see a bump in JVM heap
usage for a large dir.
I am still not satisfied with FileSystem#contentSummary(). The closest I could
find is FileSystem#getQuotaUsage() which would return number of objects in a
directory. but quota is not enabled by default.
Nits:
{code}
import static org.apache.hadoop.fs.CommonConfigurationKeysPublic.*;
{code}
TrashPolicyWithSafeDelete should not do wildcard import
{code}
LOG.debug("DIR "+ path + " in trash is too large, try safe delete.");
{code}
This is not necessarily true, if skipCheckLimit is true.
> Namenode gets stuck when deleting large dir in trash
> ----------------------------------------------------
>
> Key: HDFS-13769
> URL: https://issues.apache.org/jira/browse/HDFS-13769
> Project: Hadoop HDFS
> Issue Type: Improvement
> Affects Versions: 2.8.2, 3.1.0
> Reporter: Tao Jie
> Assignee: Tao Jie
> Priority: Major
> Attachments: HDFS-13769.001.patch, HDFS-13769.002.patch,
> HDFS-13769.003.patch, HDFS-13769.004.patch
>
>
> Similar to the situation discussed in HDFS-13671, Namenode gets stuck for a
> long time when deleting trash dir with a large mount of data. We found log in
> namenode:
> {quote}
> 2018-06-08 20:00:59,042 INFO namenode.FSNamesystem
> (FSNamesystemLock.java:writeUnlock(252)) - FSNamesystem write lock held for
> 23018 ms via
> java.lang.Thread.getStackTrace(Thread.java:1552)
> org.apache.hadoop.util.StringUtils.getStackTrace(StringUtils.java:1033)
> org.apache.hadoop.hdfs.server.namenode.FSNamesystemLock.writeUnlock(FSNamesystemLock.java:254)
> org.apache.hadoop.hdfs.server.namenode.FSNamesystem.writeUnlock(FSNamesystem.java:1567)
> org.apache.hadoop.hdfs.server.namenode.FSNamesystem.delete(FSNamesystem.java:2820)
> org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.delete(NameNodeRpcServer.java:1047)
> {quote}
> One simple solution is to avoid deleting large data in one delete RPC call.
> We implement a trashPolicy that divide the delete operation into several
> delete RPCs, and each single deletion would not delete too many files.
> Any thought? [~linyiqun]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
