activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christopher Shannon <christopher.l.shan...@gmail.com>
Subject Re: ActiveMQ Brokers starts the reindex of the KahaDB on every startup
Date Wed, 18 Nov 2015 15:18:16 GMT
Tim, I think you are referring to this fix here:
https://issues.apache.org/jira/browse/AMQ-4832

Looks like it was done in 5.9.1 and later.

There's also been some other fixes and improvements done in this area
recently so I would definitely recommend upgrading to a more recent version
and giving that a shot.

On Wed, Nov 18, 2015 at 8:55 AM, Tim Bain <tbain@alumni.duke.edu> wrote:

> I have a vague memory of seeing a fix for exactly this bug in one of the
> most recent (5.11.x or later) versions, but I couldn't find it in a quick
> search of JIRA.  Can anyone confirm my memory?
>
> If I'm remembering correctly, then an upgrade to 5.12.1 should fix your
> problem.  If your organization will let you, you could upgrade your staging
> environment to see if the observed behavior changes.
>
> Tim
> On Nov 18, 2015 2:53 AM, "patrikwettergren" <patrik.wettergren@tr.com>
> wrote:
>
> > Hello we are running on a Master/Slave Setup of two ActiveMQ Brokers with
> > version 5.9.
> >
> > We have taken a copy of our ActiveMQ Production KahaDB and placed it in
> the
> > Stage Environment. This to resolve an issue with DLQ placed Messages.
> >
> > The issue we are seeing now is that the Stage ActiveMQ Brokers are
> running
> > a
> > full re-index every time they are restarted and when the re-indexing are
> > done the Broker states that there is not enough  space on the disk and it
> > then caps the usage of the disk to what it reads is available.
> >
> > The problem is here that the used disk space is from the copied KahaDB
> that
> > the AMQ Broker just re-indexed.
> >
> > So in other word it reads the KahaDB files fine to re-index the data in
> > them, but does not take into account of what diskspace it is already
> using
> > when reading space available when it is finished the re-indexing.
> >
> > The Size of the KahaDB Set is 226GB it is placed on a 500GB Mount on and
> > the
> > ActiveMQ Brokers are runninng on Oracle Linux Server release 6.5
> > OL-6.5.1-64
> > servers with JDK 6.
> >
> >
> > In activemq.xml we have set the flow control so that anyhting above 75%
> of
> > storage usage stops messages handling on specific queues and the storage
> > usage is set :
> >    <systemUsage>
> >             <systemUsage>
> >                 <memoryUsage>
> >                     <memoryUsage percentOfJvmHeap="75" />
> >                 </memoryUsage>
> >                 <storeUsage>
> >                     <storeUsage limit="400 gb"/>
> >                 </storeUsage>
> >                 <tempUsage>
> >                     <tempUsage limit="50 gb"/>
> >                 </tempUsage>
> >             </systemUsage>
> >         </systemUsage>
> >
> > But even so the AMQ Brokers only sees the availble Disk space in '500G
> > 226G
> > 275G  46%' on the Mount and activates the flow control since the 275Gb
> > available space is smaller then the 400GB stated in activemq.xml.
> >
> > So the question is now why is the Broker doing this even after it have
> read
> > through the KahaDB data and reindexed it and updated the db.data file and
> > is
> > using the 226GB but does not take this into account  when reading the
> > 'storageUsage' limit in activemq.xml
> >
> >
> > I might be missing something here ot is it just as simple that I need to
> > get
> > the Mountpoint of 500GB larger. So any help would be very appriciated.
> >
> >
> > Kind regards
> > Patrik Wettergren
> >
> >
> >
> >
> >
> > --
> > View this message in context:
> >
> http://activemq.2283324.n4.nabble.com/ActiveMQ-Brokers-starts-the-reindex-of-the-KahaDB-on-every-startup-tp4704035.html
> > Sent from the ActiveMQ - User mailing list archive at Nabble.com.
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message