[ https://issues.apache.org/jira/browse/CASSANDRA-8160?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14330201#comment-14330201 ]
Benedict commented on CASSANDRA-8160: ------------------------------------- cat /dev/null is probably a better approach to this. If we want to do this inside of C*, we should emulate that. We should be provided with a target size of data to populate, read the readMeter statistics, and select the top ranked sstables by reads that approximately meet this size, and just iterate through their contents. No fadvise calls, since we don't want the OS to do anything special with the caching of these pages, we just want them present as a minimal startup set. Once we've done this, we then populate the key cache to touch any common pages in sstables we ranked too lowly to be prewarmed.. > CF level option to call posix_fadvise for sstables on creation and startup > -------------------------------------------------------------------------- > > Key: CASSANDRA-8160 > URL: https://issues.apache.org/jira/browse/CASSANDRA-8160 > Project: Cassandra > Issue Type: New Feature > Components: Core > Reporter: Matt Stump > Assignee: Branimir Lambov > Priority: Minor > Fix For: 2.1.4 > > Attachments: trunk-8160.txt > > > We should have a CF level configuration with will result in posix_fadvise > being called for sstables for that CF. It should be called on node startup > and for new sstables. This should be configurable per CF to allow for some > CFs to be prioritized above others. Not sure if we should use > POSIX_FADV_SEQUENTIAL or POSIX_FADV_WILLNEED. -- This message was sent by Atlassian JIRA (v6.3.4#6332)