What if we include that verbose info in toString() only if trace is
enabled? :)
On 02/20/2013 02:36 PM, Dan Berindei wrote:
On Wed, Feb 20, 2013 at 11:49 AM, Mircea Markus <mmarkus(a)redhat.com
<mailto:mmarkus@redhat.com>> wrote:
I always liked this idea of categories but never saw it at use.
Are there any projects that use this logging approach?
On 20 Feb 2013, at 09:57, Sanne Grinovero wrote:
> +1 for using categories
>
> We could even experiment combining multiple categories, for
> example in
> this case you could have a "RPCDispatcher" category and also have a
> "RPCDispatcher.includeCacheEntries" which will make descriptions
> more/less verbose.
That's not what I understand by a category - "logical process" as
defined by David. I consider "Remoting" or "Rehashing" a
category,
but RPCDIspatcher is just an entity (too fine grained)
and RPCDispatcher.includeCacheEntries even more so.
Also that wouldn't necessarily solve the problem Manik raised: in
this particular case the toString of StateResponseCommand is huge.
Adrian/Dan is this needed for debugging state transfer issues? If
so +1 for managing it with the verbose flag.
Yeah, we couldn't introduce a RpcDispatcher.includeCacheEntries
category anyway because the cache entries are included in the
command's toString() - the logger can't do anything to filter them out.
I think we could eliminate the cache entries from
StateResponseCommand.toString() (and the segment owners from
ConsistentHash.toString()), and only log them separately, under a
different category/class name.
_______________________________________________
infinispan-dev mailing list
infinispan-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev