Fixed Issues in Kudu
Review the list of Kudu issues that are resolved in Cloudera Runtime 7.3.1
- KUDU-3576: Fix the Connection timeout After Tablet Server Restart
-
In Kudu, if a Java client application maintains an open connection to a tablet server and the tablet server is restarted or encounters a network error, the client cannot re-establish communication with the tablet server even after it comes back online. The fix resolves the issue by updating the Kudu Java client.
Apache Jira: KUDU-3576
- KUDU-3496: Support for SPNEGO dedicated keytab
- Kudu now supports configuring a dedicated Spnego
keytab.
Apache Jira: KUDU-3496
- KUDU-3524: Fix crash when sending periodic keep-alive requests
- The fix ensures that Kudu clients do not crash when sending keep-alive requests.
- KUDU-3497: Optimize leader rebalancer algorithm
- Optimized the leader balancing algorithm to effectively handle corner cases detailed in the Jira.
- KUDU-3447: Limit the usage of network bandwidth of tablet copying
- Two new flags are introduced in Kudu CLI tools to copy tablets
kudu tablet copy_from_remotecommand to limit the speed of the copy task and avoid resource contention. - KUDU-3353: Add an immutable attribute to column schema
- Introduced
Immutablecolumn. It’s useful to represent a semantically constant entity. - KUDU-3351: Add insert error count metrics in WriteResponsePB
- Statistics on various write operations are now available via Kudu client API at the session level.
- KUDU-3526: Scanner should bound with a tserver in java client
- The scanner in the Kudu Java client now binds with the Kudu tablet server. This prevents scanning failures that occur when scanning from the leader replica and leadership changes to a different replica.
