Guangxu Cheng created HBASE-18804:
-------------------------------------
Summary: Sanity check the number of regions when creating a table
Key: HBASE-18804
URL: https://issues.apache.org/jira/browse/HBASE-18804
Project: HBase
Issue Type: Bug
Components: master
Affects Versions: 2.0.0, 1.5.0
Reporter: Guangxu Cheng
Assignee: Guangxu Cheng
A few days ago, a table was erroneously created with NUMREGIONS set to 200k
which resulted in the master down.
I got exceptions like:
{code:java}
2017-08-17 11:10:17,970 WARN [main-SendThread(localhost:2181)]
zookeeper.ClientCnxn: Session 0x45c9d4132c547f1 for server
localhost/localhost:2181, unexpected error, closing socket connection and
attempting reconnect
java.io.IOException: Packet len6842180 is out of range!
at
org.apache.zookeeper.ClientCnxnSocket.readLength(ClientCnxnSocket.java:112)
at
org.apache.zookeeper.ClientCnxnSocketNIO.doIO(ClientCnxnSocketNIO.java:79)
at
org.apache.zookeeper.ClientCnxnSocketNIO.doTransport(ClientCnxnSocketNIO.java:366)
at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1081)
{code}
That's because a lot of regions need to assign in a short interval, and the
size of the node /hbase/region-in-transition exceeds the default value.
We solved this problem by setting "jute.maxbuffer".
But we can sanity check the number of regions when creating a table and reject
the requests.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
