Dear HBase Community,
I'm encountering an unexpected behavior in one of our HBase tables and
would appreciate your insights. Here are the details:
Table Configuration:
NAME => 'EXAMPLE_TABLE', FAMILIES => [ { NAME => 'CF', BLOOMFILTER =>
'ROW', VERSIONS => '1', IN_MEMORY => 'false', KEEP_DELETED_CELLS => 'TTL',
DATA_BLOCK_ENCODING => 'NONE', TTL => '864000 SECONDS (10 DAYS)',
COMPRESSION => 'LZ4', MIN_VERSIONS => '0', BLOCKCACHE => 'true', BLOCKSIZE
=> '65536', REPLICATION_SCOPE => '1', ENCRYPTION => 'AES' } ]
Observed Behavior:
1. When inserting data with timestamps older than 1 year, the data
appears to be immediately marked as deleted (tombstones).
2. This behavior doesn't occur in other tables with similar
configurations.
3. A RAW scan of the table shows that the data is present, but it's not
visible in regular scans.
Expected Behavior: Data should be inserted normally, regardless of
timestamp. Deletion should only occur when explicitly deleted, and then the
deleted data should be kept for the TTL period before being removed during
major compaction.
Questions:
1. Why is data with old timestamps being treated differently?
2. Is there an interaction between TTL, KEEP_DELETED_CELLS, and old
timestamps that could cause this behavior?
3. How can I modify the table configuration or data insertion process to
handle historical data correctly?
HBase version: 1.4.14
Any insights or suggestions on how to resolve this issue or further debug
it would be greatly appreciated.
*Regards,*
*Manimekalai K*
