We recently migrated from MapR 5.2.2 to Mapr 6.0.1. We use the python library happybase (v1.1.0) to access mapr db entries via python (on a mapr client). After the upgrade, we noticed that the happybase library is not working as expected. Heres the scenario:
Sample code block:
import happybaseconnection = happybase.Connection(<host>, 9090)table = connection.table(<table_name>)table.scan().next()
Old cluster that works:
Mapr Version: 5.2.2
New cluster that does NOT work:
Mapr Version: 6.0.1
So, as I said after the upgrade, on the new cluster, the above code block just waits for eternity after the "table.scan().next()" command. On the old cluster, the same code block retrieves a row from the table specified.
This kind of behavior is also observed for "table.row()" and similar commands. The hbase thrift logs on the hbase thrift server do not contain any errors whatsoever. I do however see the following entry in the log:
INFO [thrift-worker-10] client.ConnectionFactory: ConnectionFactory receives mapr.hbase.default.db(maprdb), set clusterType(MAPR_ONLY), user(mapr), hbase_admin_connect_at_construction(false)
Also, another thing I tried out was that the hbase shell commands like "scan" and "get" works with the new version without any issues.
So, my question is, did we miss any steps in the upgrade? What seems to have cause this issue?