添加链接
link管理
链接快照平台
  • 输入网页链接,自动生成快照
  • 标签化管理网页链接

Hello everyone. So the problem is next. I have Apache Hive 2.3.6. on my cluster. I start multiple sessions simultaneously and independently from each other. And then temporary .pipeout files with directories are created and if certain session end up with an error this empty files keep accumulating because they are not deleted if the session end up not right. The similar issue was here, but it was not solved - https://issues.apache.org/jira/browse/HIVE-6091 .
So the question is how can I find the list of active sessions on my cluster or how can I check whether a session is active or not by sessionId in Hive? I need this information because I want to delete those .pipeout files and in order to do that I need to know which file are still in use (belongs to active session) and which are not! Thanks!

Thanks for the answer, but the result is negative, I don't know why, but the UI simply do not show appropriate information. I tested it, I started 2 different sessions of Hive simultaneously and initially yes - the UI shows new running sessions of Hive properly, but then when I reproduce the error of session with "kill - 9" command the UI do not updates and keep showing that there are still 2 active sessions (truly only one left - first screen ). I waited a couple minutes updating the page and then UI showed that there are no active sessions, that both are finished successfully (truly - one session is killed and one is still active and proceed queries - second screen ).

Hi @Kicker

Question: How can I find the list of active sessions on my cluster or how can I check whether a session is active or not by sessionId in Hive?

Answer: Could you please take a look at the below link where we discussed a similar question.
https://community.cloudera.com/t5/Support-Questions/How-many-users-connected-to-HiveServer2/m-p/3223...

You may have to log in to hiveserver2 and run the below command to see the active number of connections to hiveserver2.
netstat -ntpla | grep 10000 | grep -i ESTABLISHED ### Instead of 10000 you have to mention your HS2 Port(if you use http more the port number will be 10001)
netstat -ntpla | grep 10000 ### This gives you the detail of Established, Close_wait, and other processes as well.


If you want to check the Active sessions you can find the details in Hiveserver2 Web UI
Ambari > Hive > QuickLinks > Hiveserver2 WebUI

If you are happy with the reply, mark it Accept as Solution

Apache Hadoop and associated open source project names are trademarks of the Apache Software Foundation. For a complete list of trademarks, click here.