HBASE-30160: Prevent region creation if the encoded region names are …

[hingu-8103]

…the same

AI Usage

Used github copilot to implement the changes and then hand reviewed all the changes.

Testing

1. Reproduced the error for CreateTableProcedure and SplitTableRegionProcedure as mentioned in the ticket.
2. Applied the fix.
3. Tested the same scenario manually for CreateTableProcedure and SplitTableRegionProcedure.
4. Got the DoNotRetryException instead of hanging procedure for long time.
   ex.

key1 = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,94,66,185,153,219,183,152,87,250,161,224,241,188,9,104,93,49,83,91,38,117,42,147,161,38,82,122,70,135,158,151,48,132,229,185,227,108,110,42,108,7,12,239,3,150,81,189,67,33,177,222,99,45,251,43,17,131,104,193,190,36,31,174,149,175,211,87,7,138,1,250,241,186,131,140,125,165,65,49,131,174,174,248,230,249,229,70,167,201,26,254,77,236,7,222,109,14,109,158,151,244,22,8,148,168,138,56,55,7,181,118,172,231,7,16,34,252,185,31,109,189,19,86,169,237,88,240,177].pack("C*")
key2 = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,94,66,185,153,219,183,152,87,250,161,224,241,188,9,104,93,49,83,91,166,117,42,147,161,38,82,122,70,135,158,151,48,132,229,185,227,108,110,42,108,7,12,239,3,150,209,188,67,33,177,222,99,45,251,43,17,131,104,193,62,36,31,174,149,175,211,87,7,138,1,250,241,186,131,140,125,165,65,49,131,174,174,248,102,249,229,70,167,201,26,254,77,236,7,222,109,14,109,158,151,244,22,8,148,168,138,56,55,7,53,119,172,231,7,16,34,252,185,31,109,189,19,86,41,237,88,240,177].pack("C*")

hbase:037:0> create 'table1','f',SPLITS => [key1, key2]

ERROR: org.apache.hadoop.hbase.DoNotRetryIOException: Encoded region name collision detected: '495efdcd24c0fa172bef0c39ec4c9ab4' for table table1. Refusing to proceed.
	at org.apache.hadoop.hbase.util.ModifyRegionUtils.checkForEncodedNameCollisions(ModifyRegionUtils.java:110)
	at org.apache.hadoop.hbase.master.procedure.CreateTableProcedure.prepareCreate(CreateTableProcedure.java:297)
	at org.apache.hadoop.hbase.master.procedure.CreateTableProcedure.executeFromState(CreateTableProcedure.java:95)
	at org.apache.hadoop.hbase.master.procedure.CreateTableProcedure.executeFromState(CreateTableProcedure.java:60)
	at org.apache.hadoop.hbase.procedure2.StateMachineProcedure.execute(StateMachineProcedure.java:188)
	at org.apache.hadoop.hbase.procedure2.Procedure.doExecute(Procedure.java:963)
	at org.apache.hadoop.hbase.procedure2.ProcedureExecutor.execProcedure(ProcedureExecutor.java:1785)
	at org.apache.hadoop.hbase.procedure2.ProcedureExecutor.executeProcedure(ProcedureExecutor.java:1462)
	at org.apache.hadoop.hbase.procedure2.ProcedureExecutor$WorkerThread.runProcedure(ProcedureExecutor.java:2110)
	at org.apache.hadoop.hbase.trace.TraceUtil.trace(TraceUtil.java:216)
	at org.apache.hadoop.hbase.procedure2.ProcedureExecutor$WorkerThread.run(ProcedureExecutor.java:2137)

For usage try 'help "create"'

Took 0.5583 seconds

[meszibalu]

@hingu-8103 In the description of the Jira I suggested the encoded name can conflict only within a table. However it might be possible to create conflicts across multiple tables. So I can generate two special split points, one for table1,m and the other one for table2,n start keys. HBase also appends a timestamp into the hash input, so exploiting it also requires to do the split exactly at the same time (millisecond precision), but it is theoretically possible.

[hingu-8103]

@meszibalu I've addressed the concern you mentioned in the previous comment.
Thanks for pointing it out.