Zheng Hu created HBASE-22811:
--------------------------------
Summary: Deprecated the PageFilter because it's not a global page
filter but only region level page filter.
Key: HBASE-22811
URL: https://issues.apache.org/jira/browse/HBASE-22811
Project: HBase
Issue Type: Improvement
Reporter: Zheng Hu
Assignee: Zheng Hu
In HBASE-21332, one user complain about the PageFilter. Say have a table with
5 regions: (-oo, 111), [111, 222), [222, 333), [333, 444), [444, +oo), and each
region have > 10000 rows. when he use the following scan to read the table, it
will gain more than 3000 rows, that's quite confusing:
{code}
Scan scan = new Scan();
scan.withStartRow(Bytes.toBytes("111"));
scan.withStopRow(Bytes.toBytes("4444"));
scan.setFilter(new PageFilter(3000));
{code}
In fact, the state in PageFilter is region leve only, once the scan switch from
one region to another region, then we will create a new PageFilter with
rowsAccepted =0 (rows counter inside PageFilter ). So the scan will get
results as the following:
1. got 3000 rows in region [111,222), and switched to another region [222,
333);
2. got 3000 rows in region [222,333), and switched to another region [333,
444);
3. got 3000 rows in region [333,444), and reached the stopRow. stop.
Then the scan got 9000 rows in the end, that's quite confusing for HBase use.
Actually, since 1.4.x, we have the Scan.setLimit which can replace the
PageFilter now I think.
So plan to mark the PageFilter as deprecated.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
