Hello.
My current cache has 8000 entries with a total size of 24GB. On the
ext3 filesystem for this cache, there is 52493589 free inodes, so it
should be able to handle a lot more cache entries..
I just did a simple test. On my computer, it takes 14 seconds to
create 1000000 files in a single directory. Reading a single of those
files takes "no" time. Deleting them takes 10 seconds. A filesystem is
pretty fast.
$ date ; (seq 1000000 | xargs touch) ; date
Thu Aug 27 22:00:49 CEST 2009
Thu Aug 27 22:01:03 CEST 2009
$ date ; cat 12345 ; date
Thu Aug 27 22:01:29 CEST 2009
Thu Aug 27 22:01:29 CEST 2009
$ date ; (seq 1000000 | xargs rm) ; date
Thu Aug 27 22:01:50 CEST 2009
Thu Aug 27 22:02:00 CEST 2009
You should probably do your own benchmarking with test data closer to
your domain.
- Tore.
On 27. aug.. 2009, at 13.14, rushi kumar wrote:
Thanks Tore for your suggestion. I think this will not work in my
case as I
need to keep many items in the region.If you are already using this,
please
let me know how many items can be stored in disk file cache.
Thanks,
Rushi
On Wed, Aug 26, 2009 at 8:49 PM, Tore Halset <hal...@pvv.ntnu.no>
wrote:
Hello.
The disc file cache should work across different JVMs. I am still
using my
own version, but Aaron checked in a streamlined version to jcs.
Each cache item is stored as a separate file, so you should take
care that
the cache are not storing more files then the file system can handle.
Different file systems have different limitations..
https://issues.apache.org/jira/browse/JCS-58
Regards,
- Tore.
On 24. aug.. 2009, at 16.19, Aaron Smuts wrote:
I don't fully understand your message. But it sounds as if you are
trying
to share the disk cache. That cannot be done. You cannot share
the indexed
disk cache between different JVMs. It keeps the index, the keys
in memory.
. . .
If you want to share items, you have to use a lateral or remote
auxiliary.
Alternatively, you could use the JDBC disk cache.
Aaron
--- On Thu, 8/20/09, rushi kumar <rushikum...@gmail.com> wrote:
From: rushi kumar <rushikum...@gmail.com>
Subject: Configure JCS in horizontally clustered environment
To: jcs-users@jakarta.apache.org
Date: Thursday, August 20, 2009, 10:35 PM
I have configure JCS in my
application and deployed on horizontally
clustered environment. As per this, it contains multiple
physical machines
(nodes). The cluster-enabled application is deployed to
each node and is
available for requests. A load balancer provides workload
management by
distributing requests to the various nodes.
I am using indexed disk cache in the aplication and the *
jcs.auxiliary.DC.attributes.DiskPath=/Rootdir/Test/cache*
is shared by both
the physical machine. Currently the cached objects are not
being shared by
the both the nodes. It looks as if the caching is working
properly on one
physical server , but the cache is not updated on the
second physical
server.
I would like to know if there's a way to configure JCS to
run on
horizontally clustered environment.
My apologizes if these questions have already been posted
and answered.
---------------------------------------------------------------------
To unsubscribe, e-mail: jcs-users-unsubscr...@jakarta.apache.org
For additional commands, e-mail: jcs-users-h...@jakarta.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: jcs-users-unsubscr...@jakarta.apache.org
For additional commands, e-mail: jcs-users-h...@jakarta.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: jcs-users-unsubscr...@jakarta.apache.org
For additional commands, e-mail: jcs-users-h...@jakarta.apache.org
