[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17721964#comment-17721964 ] Dheeren Beborrtha commented on HDFS-12737: -- We are observing this issue in Hbase cluster of

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16788409#comment-16788409 ] Wei-Chiu Chuang commented on HDFS-12737: We saw this issue coming up again with a Spark job and

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16250627#comment-16250627 ] Todd Lipcon commented on HDFS-12737: Not following what you mean by "implement multiplexing in the

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16243215#comment-16243215 ] Yongjun Zhang commented on HDFS-12737: -- Thanks a lot [~jnp] and [~tlipcon]! I did some study and

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16235221#comment-16235221 ] Todd Lipcon commented on HDFS-12737: In the data transfer protocol we just pass tokens with each

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16235182#comment-16235182 ] Jitendra Nath Pandey commented on HDFS-12737: - [~yzhangal], The block token is also being used

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16234752#comment-16234752 ] Yongjun Zhang commented on HDFS-12737: -- Many thanks [~jnp], that make sense! If we could make the

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16227678#comment-16227678 ] Jitendra Nath Pandey commented on HDFS-12737: - BlockTokenSelector uses only Token-Kind to

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16226008#comment-16226008 ] Yongjun Zhang commented on HDFS-12737: -- Thanks [~jojochuang] for working on this issue. Hi [~jnp]

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16223009#comment-16223009 ] Wei-Chiu Chuang commented on HDFS-12737: I don't understand why the client needs to make a dummy

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16222994#comment-16222994 ] Wei-Chiu Chuang commented on HDFS-12737: Correct. This issue (new connection for every call) only

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16222989#comment-16222989 ] Kihwal Lee commented on HDFS-12737: --- You quoted {{ClientDatanodeProtocol}}, so I guess they are the

[jira] [Commented] (HDFS-12737) Thousands of sockets lingering in TIME_WAIT state due to frequent file open operations

[ https://issues.apache.org/jira/browse/HDFS-12737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16222981#comment-16222981 ] Kihwal Lee commented on HDFS-12737: --- If the same user (i.e. the same UGI instance) talks to the same