mina-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Latorre <dvl...@gmail.com>
Subject Re: [MINA 3.0] Debugging toolkit
Date Tue, 02 Feb 2010 10:45:53 GMT
2010/1/25 Alan D. Cabrera <list@toolazydogs.com>:
> On Jan 25, 2010, at 8:21 AM, Ashish wrote:
>> On Mon, Jan 25, 2010 at 9:48 PM, Emmanuel Lecharny <elecharny@gmail.com>
>> wrote:
>>> Alan D. Cabrera a écrit :
>>>> On Jan 19, 2010, at 7:41 AM, Ashish wrote:
>>>>> Had picked this form one the discussion threads and added to the MINA
>>>>> 3.0 design page, so that we don't loose this.
>>>>> had some initial thoughts around this, like we can some kind of
>>>>> visualization tool build around MINA, which can help us see/debug
>>>>> the MINA based System. One can view number of IoSession's, IoChain's,
>>>>> configuration etc.
>>>>> It could be build as an Eclipse plugin or a standalone UI talking to
>>>>> MINA code via JMX.
>>>>> I am not sure, but we can see if we can extend ProtoTiger as a
>>>>> debugging/visualization tool.
>>>>> These are just some initial thought, will refine the understanding as
>>>>> we progress further with 3.0 design.
>>>>> Any suggestion?
>>>> I wouldn't classify these as debugging toolkit but a management toolkit;
>>>> very critical.
>>>> I was also thinking about the logging discussion that we had earlier and
>>>> I
>>>> think that I've come around to the thinking that it's important to
>>>> provide
>>>> some kind of logging mechanism who's scope is narrowed to a particular
>>>> session.  (I'm still not sold on the usefulness of a general logging
>>>> filter
>>>> such as the one that we have now).
>>> With JMX, and a decent tool (something like MX4J :
>>> http://mx4j.sourceforge.net/), providing some graphs should be a piece of
>>> cake.
>> Emm ! You read my mind. I had something similar in my mind :-)
>> More like number of session, and if possible the chain associated with
>> the same..
>> there could be more such useful stuff on that UI
> It would be good if we had some kind of conventions, e.g. naming and event,
> that other protocols would follow.
>>> Being able to track down one single session is also something that a JMX
>>> based tool can do. Logs ? That's for post-mortem debugging :/
> We use logs extensively when working on production issues where sessions can
> have the lifetime of subatomic particles.  We should carefully consider the
> usefulness of logging in this toolkit.

Logging at the session level should be pretty easy with NDC/MDC,
right? The problem is of course, if their logging implementation
doesn't support these.

> Regards,
> Alan

View raw message