Support Questions

Find answers, ask questions, and share your expertise
Announcements
Celebrating as our community reaches 100,000 members! Thank you!

hiveserver2 error

avatar
Contributor

Hi experts,

Many of the devs are getting the following error when trying to connect with squirrel:

Error: org.apache.thrift.transport.TTransportException: java.net.SocketTimeoutException: Read timed out

SQLState:  08S01

ErrorCode: 0

 

I believe the issue may be because there are too many concurrent hive connections.

 

Can someone help suggest how should i configure this to allow more hive connections?

Also I find it weird that I can connect through beeline using the edge node of that cluster but Squirrel is giving an error.

 

Thanks,

Any help is much appreciated.

12 REPLIES 12

avatar
Cloudera Employee

@ryu , as a general rule of thumb Cloudera recommends that you determine the total number of HS2 servers on a cluster by dividing the expected maximum number of concurrent users on a cluster by 40. For example, if 400 concurrent users are expected, 10 HS2 instances should be available to support them.

 

Also, here are some HS2 tuning best practices that you can visit => https://docs.cloudera.com/documentation/enterprise/6/6.3/topics/admin_hive_tuning.html#hs2_perf_best...

 

Regards,

Aditya

avatar
Contributor

yeah we currently have 2 HS2 instances.

For some reason our production seems to be working fine with Squirrel.

Our dev seems to be timing out after running simple queries such as "show databases".

Beeline seems to work fine on our dev cluster.

The only difference I can think of is that our dev cluster has an external mysql server whereas the production cluster, mysql server is installed on one of the nodes.

 

Am I missing some squirrel drivers or something? Wondering why it is just squirrel that seems to have issues running queries against our dev hiveserver2.

 

Any help is much appreciated.


Thanks,

avatar
Contributor

Also Squirrel seems to be connecting to the dev cluster.

It just times out when running a query such as "show databases".

 

If squirrel stays connected for a long time, I noticed that the query will eventually return results instead of timing out.

 

Per cloudera "https://docs.cloudera.com/documentation/enterprise/latest/topics/cdh_ig_hive_metastore_configure.htm...

It says that minimum 4 dedicated cores to HS2 and 4 for hive metastore.

The server that hosts hs2 and metastore only has a total of 8 cores.

Can this be a reason for the performance issue?

Any help on this is much appreciated.

 

Thanks,