[jira] [Commented] (AMQ-6938) KahaDB's ignoreMissingJournalFiles argument doesn't actually ignore missing journal files

["Gary Tully (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/AMQ-6938?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16461059#comment-16461059 ] 

Gary Tully commented on AMQ-6938:
---------------------------------

note: if journal files are manually removed, then the index needs to be rebuild, otherwise there will be references in the index that won't exist and that is expected to error out.

removing the db.data will force an index rebuild from the available set of journal files.

the ignoreMissingJournalFiles relates to finding corrupt journal files.... it should be called ignoreCorruptJournalFiles i think. That needs some investigation to figure the best course of action. Fix of document

> KahaDB's ignoreMissingJournalFiles argument doesn't actually ignore missing journal files
> -----------------------------------------------------------------------------------------
>
>                 Key: AMQ-6938
>                 URL: https://issues.apache.org/jira/browse/AMQ-6938
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: KahaDB
>    Affects Versions: 5.13.1
>            Reporter: Tim Bain
>            Priority: Major
>
> It was reported on the user mailing list ([http://activemq.2283324.n4.nabble.com/Re-failed-to-start-ActiveMQ-td4737631.html)] that a user is seeing the broker fail to start due to "java.io.IOException: Could not locate data file /data/kahadb/db-531.log" even though ignoreMissingJournalFiles is set to true. Presumably the purpose of that option is to allow the broker to ignore missing journal files (otherwise we chose poorly when naming the option and then failed to document it adequately on the wiki), which means the option doesn't work properly in all situations.
> Looking at the code on master as of 3/25/2018, org.apache.activemq.store.kahadb.MessageDatabase.recover() calls journal.getNextLocation(null) on line 677, which will throw the exception in question when DataFile.getDataFile() is called. This call to journal.getNextLocation() occurs outside of the while loop, whereas the ignoreMissingJournalFiles value is only checked in the catch block within the while loop, so we do not consider the ignoreMissingJournalFiles value in the call to getNextLocation(), even though missing journal files can cause getNextLocation() to fail.
> Note that the file is missing because the user manually deleted it in an effort to work around AMQ-6931, but that should not be relevant to the behavior that's being seen.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)