I see ... The point is it seems not cleanup until a restart.
I put a log into DataFileAccessorPool.java to see the disposeUnused method. Then i sent a lot of messages, until my data files look like this: ls -lh data/localhost/journal/ total 50M -rw-r--r-- 1 root root 32M 2007-10-11 18:21 data-3 -rw-r--r-- 1 root root 32M 2007-10-11 18:31 data-4 -rw-r--r-- 1 root root 2.1K 2007-10-11 18:31 data-control and ls -lh data/localhost/kr-store/data/ total 40M -rw-r--r-- 1 root root 949 2007-10-11 17:57 data-blob-1 -rw-r--r-- 1 root root 3.3K 2007-10-11 17:59 data-kaha-1 -rw-r--r-- 1 root root 3.2M 2007-10-11 18:31 data-topic-acks-1 -rw-r--r-- 1 root root 6.1M 2007-10-11 18:31 data-topic-data-1 -rw-r--r-- 1 root root 17M 2007-10-11 18:31 data-topic-subs-references-1 -rw-r--r-- 1 root root 0 2007-10-11 17:57 hash-index-blob_topic___ActiveMQ.Agent-Subscriptions -rw-r--r-- 1 root root 33K 2007-10-11 17:57 hash-index-blob_topic___auction.admin-Subscriptions -rw-r--r-- 1 root root 33K 2007-10-11 17:57 hash-index-blob_topic___auction.negotiation-Subscriptions -rw-r--r-- 1 root root 17K 2007-10-11 17:57 hash-index-blob_topic___auction.notification-Subscriptions -rw-r--r-- 1 root root 0 2007-10-11 17:59 hash-index-blob_topic___auction.response-Subscriptions -rw-r--r-- 1 root root 0 2007-10-11 17:57 hash-index-topic-data_topic___ActiveMQ.Agent -rw-r--r-- 1 root root 17M 2007-10-11 18:31 hash-index-topic-data_topic___auction.admin -rw-r--r-- 1 root root 0 2007-10-11 17:57 hash-index-topic-data_topic___auction.negotiation -rw-r--r-- 1 root root 0 2007-10-11 17:57 hash-index-topic-data_topic___auction.notification -rw-r--r-- 1 root root 0 2007-10-11 17:59 hash-index-topic-data_topic___auction.response -rw-r--r-- 1 root root 510 2007-10-11 17:59 index-blob -rw-r--r-- 1 root root 1.1K 2007-10-11 17:59 index-kaha -rw-r--r-- 1 root root 2.4M 2007-10-11 18:31 index-topic-acks -rw-r--r-- 1 root root 2.4M 2007-10-11 18:31 index-topic-data -rw-r--r-- 1 root root 4.4M 2007-10-11 18:31 index-topic-subs-references -rw-r--r-- 1 root root 0 2007-10-11 17:57 lock I thought data-3 should be removed, am i right? But my log says: [DEBUG][ActiveMQ Scheduler]2007-10-11 18:32:11,381 DataFileAccessorPool:112 - file:data-4 number = 4 , length = 20506094 refCount = 89444 pool.isUsed ():true [DEBUG][ActiveMQ Scheduler]2007-10-11 18:32:11,381 DataFileAccessorPool:112 - file:data-3 number = 3 , length = 33554297 refCount = 108283 pool.isUsed ():true [DEBUG][ActiveMQ Scheduler]2007-10-11 18:32:41,385 DataFileAccessorPool:112 - file:data-4 number = 4 , length = 21730205 refCount = 94782 pool.isUsed ():true [DEBUG][ActiveMQ Scheduler]2007-10-11 18:32:41,385 DataFileAccessorPool:112 - file:data-3 number = 3 , length = 33554297 refCount = 108283 pool.isUsed ():true [DEBUG][ActiveMQ Scheduler]2007-10-11 18:33:11,382 DataFileAccessorPool:112 - file:data-4 number = 4 , length = 22957070 refCount = 100132 pool.isUsed ():true [DEBUG][ActiveMQ Scheduler]2007-10-11 18:33:11,383 DataFileAccessorPool:112 - file:data-3 number = 3 , length = 33554297 refCount = 108283 pool.isUsed ():true if pool.isUsed() is true nothing is cleanup. It seems that the file into the pool never get unused. Thanks. Att, Márcio Moraes 2007/10/11, Rob Davies <[EMAIL PROTECTED]>: > > that still looks ok - the default data file length is 32mb - I'll > expose the property to make that configurable > > > cheers, > > Rob > > http://rajdavies.blogspot.com/ > > > > On Oct 11, 2007, at 8:56 PM, Márcio Moraes wrote: > > > I build a new version from SNAPSHOT (revision 583595) and run again my > > tests. > > > > After sent ~15k messages my data dir was 34mb > > > > I wait 10 minutes but data still with 34mb ... > > > > then i restart activemq and data dir was now with 20mb > > > > Is this a ok behavior? How log it takes to clean up? Always do i need > > restart? > > > > Thanks, > > > > Márcio Moraes > > > > 2007/10/11, Márcio Moraes <[EMAIL PROTECTED]>: > >> > >> Ok, I will try and let you know ... > >> > >> Thanks. > >> > >> 2007/10/11, Rob Davies <[EMAIL PROTECTED]>: > >>> > >>> ah - in that case - please use a later snapshot - there was a bug > >>> with selectors which is now fixed > >>> > >>> > >>> cheers, > >>> > >>> Rob > >>> > >>> http://rajdavies.blogspot.com/ > >>> > >>> > >>> > >>> On Oct 11, 2007, at 6:59 PM, Márcio Moraes wrote: > >>> > >>>> I think messages have been readed. > >>>> > >>>> Because i have 2 durable consumers with selector. > >>>> > >>>> The first consumer/selector accept the messages the other one does > >>>> not. > >>>> > >>>> I think message should be considered readed even to the second > >>>> consumer. > >>>> > >>>> Am i right? > >>>> > >>>> Att, > >>>> > >>>> Márcio Moraes > >>>> > >>>> > >>>> 2007/10/11, Rob Davies < [EMAIL PROTECTED]>: > >>>>> > >>>>> Hi Marcio, > >>>>> > >>>>> the messages are stored in a journal file log - the whole log > >>>>> will be > >>>>> removed when all the messages have been read. This is done for > >>>>> performance > >>>>> > >>>>> > >>>>> cheers, > >>>>> > >>>>> Rob > >>>>> > >>>>> http://rajdavies.blogspot.com/ > >>>>> > >>>>> > >>>>> > >>>>> On Oct 11, 2007, at 2:59 PM, Márcio Moraes wrote: > >>>>> > >>>>>> Hi, > >>>>>> > >>>>>> I'm testing ActiveMQ-5.00-RC1 under the following scenario: > >>>>>> > >>>>>> - Sending N messages to a Durable Topic > >>>>>> - Two consumers identified with its clientId > >>>>>> - Both has a selector and only one can accept the test messages > >>>>>> > >>>>>> All works fine, except the messages are never removed from the > >>>>>> data > >>>>>> directory, it grows infinitely. > >>>>>> > >>>>>> Is it a bug? > >>>>>> > >>>>>> My work around may be cron a script to clean that. > >>>>>> > >>>>>> Att, > >>>>>> > >>>>>> Márcio Moraes > >>>>> > >>>>> > >>> > >>> > >> > >
