[ https://jira.jboss.org/jira/browse/JBMESSAGING-1678?page=com.atlassian.ji... ] Clebert Suconic commented on JBMESSAGING-1678: ---------------------------------------------- This is because the method load has two lists: final Set<Long> recordsToDelete = new HashSet<Long>(); final List<RecordInfo> records = new ArrayList<RecordInfo>(); The method load will add to both recordsToDelete and records. At the end of the method, it will only consider the records that were not deleted: for (RecordInfo record : records) { if (!recordsToDelete.contains(record.id)) { committedRecords.add(record); } } But In a scenario where compacting is not working, records and recordsToDelete will grow too large, up to the point it won't fit on the memory in a single load. We could just make compacting mandatory, or change the loading method to use some sort of Map. It's a bit of a challange here as the journal needs to keep the adds and update records in order. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/jira/browse/JBMESSAGING-1678?page=com.atlassian.ji... ] Clebert Suconic commented on JBMESSAGING-1678: ---------------------------------------------- That's not what I meant. As far as I'm in concern, the updates needs to be in order... if you did Add,1 Add2, Update1 Update2 I need to load them at the same order. as the Update could be a transfer to a DLQ, redelivery.. etc. If I used some ordered map, the result at the end would be: Add1, Update1 Add2, Update2 Unless you tell me this would be ok. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/jira/browse/JBMESSAGING-1678?page=com.atlassian.ji... ] Clebert Suconic closed JBMESSAGING-1678. ---------------------------------------- Resolution: Done I have committed a fix, and added a stress test to validate this issue. I didn't want to change any ordering on the journal, so I added a cleanup on every X records as discussed on the forums. We will revisit this issue if this is not acceptable. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira