[
https://issues.apache.org/jira/browse/PHOENIX-2524?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Josh Elser updated PHOENIX-2524:
--------------------------------
Attachment: PHOENIX-2524.01.patch
Ok, here's a new patch which also fixes a bunch of other things I found along
the way running pherf against PQS.
* Jars not in tarball (the original issue)
* Very wrong timings logger in pre-load/write path
* Use "jdbc:phoenix:thin:" when appropriate CLI options are present
* Fix some result CSV headers for PQS
* Avoid repeatedly creating a PreparedStatement for every row.
With this, depending on the width of the table, I'm seeing performance of PQS
between 2x-3x slower than the normal thick client. The wider the table, the
less the extra HTTP call affects us (we amortize the cost a bit when we start
shipping more data in the call).
> phoenix-pherf jars not included in binary tarball
> -------------------------------------------------
>
> Key: PHOENIX-2524
> URL: https://issues.apache.org/jira/browse/PHOENIX-2524
> Project: Phoenix
> Issue Type: Bug
> Reporter: Josh Elser
> Assignee: Josh Elser
> Fix For: 4.7.0
>
> Attachments: PHOENIX-2524.01.patch, PHOENIX-2524.patch
>
>
> Tried to run the {{pherf-standalone.py}}, got a message that it couldn't load
> the class {{org.apache.phoenix.pherf.Pherf}}.
> Looking at the classpath string, it never found the pherf jar. This is
> because it wasn't bundled in {{lib}}.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
