[
https://issues.apache.org/jira/browse/ACCUMULO-1892?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Eric Newton updated ACCUMULO-1892:
----------------------------------
Description:
examples.simple.RandomBatchWriter takes a command line arg for how many rowids
to write.
In the functional tests, this arg is used to coordinate a number of expected
reads with the examples.simple.RandomBatchReader example.
the reader implementation ensures that it generates the passed number of query
rows, but the write just makes num attempts at creating random rowids.
For a small number of goal rows and a large range (like the examples used for
testing bloom filter speeds early in the examples integration test), this works
fine.
For a larger number of goal rows (like the examples used to just write/read 10k
rows later in the examples integration test), collisions in the selected row
ids will cause the writer to actually generate fewer than the expected number.
That, in turn, will cause the attempted read integration test to fail.
was:
examples.simple.RandomBatchWriter takes a command line arg for how many rowids
to write.
In the functional tests, this arg is used to coordinate a number of expected
reads with teh examples.simple.RandomBatchReader example.
the reader implementation ensures that it generates the passed number of query
rows, but the write just makes num attempts at creating random rowids.
For a small number of goal rows and a large range (like the examples used for
testing bloom filter speeds early in the examples integration test), this works
fine.
For a larger number of goal rows (like the examples used to just write/read 10k
rows later in the examples integration test), collisions in the selected row
ids will cause the writer to actually generate fewer than the expected number.
That, in turn, will cause the attempted read integration test to fail.
> examples.simple.RandomBatchWriter might not write the specified number of
> rowids
> --------------------------------------------------------------------------------
>
> Key: ACCUMULO-1892
> URL: https://issues.apache.org/jira/browse/ACCUMULO-1892
> Project: Accumulo
> Issue Type: Bug
> Components: docs, test
> Affects Versions: 1.4.0, 1.4.1, 1.4.2, 1.4.3, 1.4.4, 1.5.0
> Reporter: Sean Busbey
> Assignee: Sean Busbey
> Priority: Minor
> Fix For: 1.4.5, 1.5.1, 1.6.0
>
> Attachments: ACCUMULO-1892.1.patch.txt, ACCUMULO-1892.2.patch.txt,
> ACCUMULO-1892.3.patch.txt
>
>
> examples.simple.RandomBatchWriter takes a command line arg for how many
> rowids to write.
> In the functional tests, this arg is used to coordinate a number of expected
> reads with the examples.simple.RandomBatchReader example.
> the reader implementation ensures that it generates the passed number of
> query rows, but the write just makes num attempts at creating random rowids.
> For a small number of goal rows and a large range (like the examples used for
> testing bloom filter speeds early in the examples integration test), this
> works fine.
> For a larger number of goal rows (like the examples used to just write/read
> 10k rows later in the examples integration test), collisions in the selected
> row ids will cause the writer to actually generate fewer than the expected
> number. That, in turn, will cause the attempted read integration test to
> fail.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
