Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
HBase >> mail # user >> Region is in transition, in CLOSING state

Copy link to this message
Region is in transition, in CLOSING state

We are using hbase-0.90.4, and after one of our region servers went down
master reported that one region is in transition, in CLOSING state:

INFO org.apache.hadoop.hbase.master.AssignmentManager: Regions in
transition timed out:  access_tokens,,1312452518317.1878726708
state=CLOSING, ts=1328975812113
INFO org.apache.hadoop.hbase.master.AssignmentManager: Region has been
CLOSING for too long, this should eventually complete or the server will
expire, doing nothing

Region server that went down was hosting 8 regions and all
were successfully closed and  reassigned to  other region servers.
Problem appeared when master tried to rebalance cluster by moving regions
from overloaded servers to less loaded servers:

INFO org.apache.hadoop.hbase.master.LoadBalancer: Calculated a load balance
in 0ms. Moving 3 regions off of 2 overloaded servers onto 2 less loaded

And again 2 regions were successfully moved to less loaded servers
but access_tokens,,1312452518317.1878726708 region stucked in CLOSING
state. We resolved issue by restarting cluster (we had to kill master).

My question is what is best approach in situations like this ? (i have
tried manual unassign from shell did not work)
I have seen few bugs that are addressing similar issue, is this fixed in