activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Bain (JIRA)" <j...@apache.org>
Subject [jira] [Created] (AMQ-5547) Provide a means to compact sparsely-populated KahaDB data files
Date Wed, 28 Jan 2015 17:34:35 GMT
Tim Bain created AMQ-5547:
-----------------------------

             Summary: Provide a means to compact sparsely-populated KahaDB data files
                 Key: AMQ-5547
                 URL: https://issues.apache.org/jira/browse/AMQ-5547
             Project: ActiveMQ
          Issue Type: Improvement
          Components: Message Store
            Reporter: Tim Bain


AMQ-5542 describes a bug in our handling of KahaDB files where most but not all of the messages
have been removed.  In the comments, we describe a scenario where many files could be kept
simply because of a single message in an older KahaDB file, because ack messages in file N+1
apply to messages in file N and so N+1 must be kept as long as N is.

The problem exists because KahaDB data files can't be compacted, which is why mKahaDB (multi-KahaDB)
was developed.  We should either use mKahaDB, or we should find some other mechanism to ensure
that unneeded data files aren't kept around simply because a single old message is still alive.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message