[
https://issues.apache.org/jira/browse/CASSANDRA-2701?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13279357#comment-13279357
]
sankalp kohli commented on CASSANDRA-2701:
------------------------------------------
ON reading through the startup code in CassandraDaemon, it looks like having
memcache as a provider will not create coherence problems.
When we are creating Table objects for all keyspaces, we initialize all row
cache for all CF for each keyspace. This happens from this call.
// initialize keyspaces
for (String table : Schema.instance.getTables())
{
if (logger.isDebugEnabled())
logger.debug("opening keyspace " + table);
Table.open(table);
}
Table.open has this line
for (ColumnFamilyStore cfs : tableInstance.getColumnFamilyStores())
cfs.initRowCache();
And then later during startup, we do a replay of commit log.
So if there is any data in commit log, it will be applied and hence will make
it to memcache.
It will work the same way if we are persisting the row cache keys on disk in
saved_cache folder.
> Memcache Row Cache Provider
> ---------------------------
>
> Key: CASSANDRA-2701
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2701
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Chris Goffinet
> Priority: Trivial
>
> Create a row cache provider that uses memcached. We would like to contribute
> our provider we wrote that we use in production. We co-locate memcached on
> every node, and utilize Cassandra's ability to do replication + routing.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
