cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jonathan Ellis (JIRA)" <>
Subject [jira] Updated: (CASSANDRA-946) Add a configuration and implementation to populate the data into memory
Date Sun, 11 Apr 2010 02:34:41 GMT


Jonathan Ellis updated CASSANDRA-946:

    Attachment: 946-v4.txt

Committed the "fix utils" patch separately.  Removed old patch versions to reduce confusion.

Attached v4 with some cleanup, and renaming to "preloadRowCache."  Changes to CompactionManager
were unnecessary since compacting doesn't actually change the data involved, so removed that.
 Load from the CFS constructor instead the way you had it originally.

With the removal of the compaction code, preloadRowCache is never called anywhere.

Checking metadata.rowCacheSize == 1 is broken since this is the absolute size of the cache,
not the configuration size.  You'll have to preserve the original configuration string somewhere
to check.

This adds the preload cache option to the Thrift call, but doesn't update system_add_column_family.
 Which is why it fails to compile...

I shouldn't be reviewing patches that haven't been compiled, let alone tested.  Please attach
screenshots of your testing to the next one.

> Add a configuration and implementation to populate the data into memory
> -----------------------------------------------------------------------
>                 Key: CASSANDRA-946
>                 URL:
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>    Affects Versions: 0.7
>         Environment: Linux, Cassandra
>            Reporter: Vijay
>            Assignee: Vijay
>            Priority: Minor
>             Fix For: 0.7, 0.8
>         Attachments: 946-Add_Config_v003.txt, 946-Documentation_v002.txt, 946-Load_Data_v003.txt,
946-v4.txt, 946_Fix_Utils_v003.txt, screenshot.png
> Proactively load data into the memory when the node is started, there will be a configuration
to enable this function and will be per Columnfamily. The requirement is to speed up the reads
for data which can reside 100% of in the memory.... In addition to enabling the RowCache to
100% we can do this so as upgrades or any other means of restart will not clear the cache
in the server.

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators:
For more information on JIRA, see:


View raw message