One problem about split

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Hyperspace master and hypertable master are in different machines. And 
three range servers are configured.
2. Start hyperspace master, hypertable master and three range servers.
3. Create table and load data as the followings:
hypertable> create table Pages ( date, "refer-url", "http-code");
hypertable> load data infile ROW_KEY_COLUMN=rowkey "access.tsv" into table 
Pages;
hypertable> load data infile ROW_KEY_COLUMN=rowkey "access.tsv" into table 
Pages;
hypertable> select * from METADATA;
0:    StartRow        0:
0:    Location        192.168.1.16_38060
1:events.themonitor.com/harlingen-tx/venues/show/235367-club-cielo      
Files:default   /hypertable/tables/Pages/default/AB2A0D28DE6B77FFDD6C72AF/c
s0;

1:events.themonitor.com/harlingen-tx/venues/show/235367-club-cielo      
StartRow
1:events.themonitor.com/harlingen-tx/venues/show/235367-club-cielo      
Location        192.168.1.16_38060
1:    
Files:default   /hypertable/tables/Pages/default/AB2A0D28DE6B77FFDD6C72AF/c
s0;

1:    StartRow        events.themonitor.com/harlingen-
tx/venues/show/235367-club-cielo
1:    StartRow
1:    Location        192.168.1.18_38060 

What is the expected output? What do you see instead?
Split should not occur because cs0 is only 4.5M.

What version of the product are you using? On what operating system?
hypertable-0.9.0.7-alpha on linux.

Please provide any additional information below.

Original issue reported on code.google.com by lanlan.y...@gmail.com on 31 Jul 2008 at 7:55

[GoogleCodeExporter]

I think this is expected behavior. Hypertable uses a dynamic split threshold to
increase throughput.

Original comment by vic...@gmail.com on 31 Jul 2008 at 9:59

[GoogleCodeExporter]

Yes, the split threshold starts out very small and then doubles on each split 
until it reaches the configured 
*.Range.SplitSize value.

Original comment by nuggetwh...@gmail.com on 1 Apr 2010 at 5:11

• Changed state: Invalid