[HBASE-7504] -ROOT- may be offline forever after FullGC of RS

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.94.3
• Fix Version/s: 0.94.5, 0.95.0
• Hadoop Flags: Reviewed
• Assignee: Chunhui Shen
• Reporter: Chunhui Shen
• Created: 07/Jan/13 09:42
• Updated: 26/Feb/13 08:27
• Resolved: 11/Jan/13 05:29

Description

1.FullGC happen on ROOT regionserver.
2.ZK session timeout, master expire the regionserver and submit to ServerShutdownHandler
3.Regionserver complete the FullGC
4.In the process of ServerShutdownHandler, verifyRootRegionLocation returns true
5.ServerShutdownHandler skip assigning ROOT region
6.Regionserver abort itself because it reveive YouAreDeadException after a regionserver report
7.ROOT is offline now, and won't be assigned any more unless we restart master

Master Log:

2012-10-31 19:51:39,043 DEBUG org.apache.hadoop.hbase.master.ServerManager: Added=dw88.kgb.sqa.cm4,60020,1351671478752 to dead servers, submitted shutdown handler to be executed, root=true, meta=false
2012-10-31 19:51:39,045 INFO org.apache.hadoop.hbase.master.handler.ServerShutdownHandler: Splitting logs for dw88.kgb.sqa.cm4,60020,1351671478752
2012-10-31 19:51:50,113 INFO org.apache.hadoop.hbase.master.handler.ServerShutdownHandler: Server dw88.kgb.sqa.cm4,60020,1351671478752 was carrying ROOT. Trying to assign.
2012-10-31 19:52:15,939 DEBUG org.apache.hadoop.hbase.master.ServerManager: Server REPORT rejected; currently processing dw88.kgb.sqa.cm4,60020,1351671478752 as dead server
2012-10-31 19:52:15,945 INFO org.apache.hadoop.hbase.master.handler.ServerShutdownHandler: Skipping log splitting for dw88.kgb.sqa.cm4,60020,1351671478752

No log of assigning ROOT

Regionserver log:

2012-10-31 19:52:15,923 WARN org.apache.hadoop.hbase.util.Sleeper: We slept 229128ms instead of 100000ms, this is likely due to a long garbage collecting pause and it's usually bad, see http://hbase.apache.org/book.html#trouble.rs.runtime.zkexpired