HI, everybody, and Cloudera support:
as we know, there are no any HIVE HA solution for Cloudera Hadoop, it means there is just one HIVE service. so what i am considering is if this HIVE service is down, how to do it. currently, i set the configuration of hive restart if HIVE service failed. but what can we do the server is down?
my solution is to use HAproxy to resolve this kind issue. and it works fine right now. but the problems is if i enable kerberos for Cloudera Hadoop, then i will failed. the errors is GSS init failed. this is easy for me to understand, since
HUE send the request to HAPROXY server, but HAPROXY server is not real server, so it will transfer this request to HIVE server, that's why the GSS init failed.
i have read many times about Impala HA by Haproxy, but how to do it for HIVE ? i didn't find any introduction on it
This thread seems to be old and unanswered for a long time, but for latest CDH and CM, you can enable using HAProxy. See doc below for details:
So it is simply setup HAProxy to load balance HS2 and then tell CM to use it. If it is kerberized, CM does the keytab files setup for you as well.