Hi, I am using Apache Phoenix version 4.3.1 with HBase 0.98.11-hadoop2. I am using the following command to split my table:
create table if not exists "JAIA" (seq unsigned_int not null, epug.cid unsigned_int, epug.e_id unsigned_smallint, epug.k VARBINARY, epug.pkt_ts unsigned_long, epug.top_idx unsigned_int, epug.type unsigned_tinyint, epug.v VARBINARY CONSTRAINT pk PRIMARY KEY (seq)) immutable_rows=true,default_column_familly='epug',epug.VERSIONS=1,COMPRESSION='SNAPPY' split on(10000000,20000000,30000000,40000000,50000000,60000000,70000000,80000000,90000000,100000000,110000000,120000000,130000000,140000000,150000000,160000000,170000000,180000000,190000000,200000000,210000000,220000000,230000000,240000000,250000000,260000000,270000000,280000000,290000000,300000000,310000000,320000000,330000000,340000000,350000000,360000000,370000000,380000000,390000000,400000000,410000000,420000000,430000000,440000000,450000000,460000000,470000000,480000000,490000000,500000000,510000000,520000000,530000000,540000000,550000000,560000000,570000000,580000000,590000000,600000000,610000000,620000000,630000000,640000000,650000000,660000000,670000000,680000000,690000000) I noticed that the first region gets all the data store files. When I looked at HBase region split information, I noticed that the high bit for the end key first split is set to 1. I.e, the keys for the first table are: Start Key: blank End Key: \x80\x98\x96\x80 The hex for the first region (10M) is: 0x989680. I.e., a leading 0x80 has been added to the value which then causes the first region includes all the values that I have. For the next region, the start key has the high bit set: Start Key:\x80\x98\x96\x80 End Key:\x811-\x00 Again, not, that instead of 0x989680, the value is 0x80989680. Is this a bug? Has anybody seen this? Thanks, Behdad
