james-server-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Timothy Prepscius <timprepsc...@gmail.com>
Subject Re: overhead of JmxServerProbe?
Date Thu, 21 Jun 2012 16:44:54 GMT
Yes, that was strange.
Ok.  I have not been able to get the timeout error again.

I believe yesterday when I tested with the james server restart, I may have restarted user-adder.
I can't other wise explain why it worked.


Today I created a user.  (which worked without timeout after 12 hours).
I immediately killed james and restarted it 10 seconds later.  (sockets)
Attempted to create another user and got a jmx exception.


So.  At this point I believe my e-mail from yesterday gave incorrect information.


--

At this point, I would guess that:

1.  I remember the initial exception incorrectly, and that I never got a timeout.  I could
have just gotten a connection problem, and I mentally inserted the word timeout.

2. If there was in fact a timeout, perhaps I had killed the James server, and somehow, as
the connection timed out, and tried to reconnect it caught this exception and attributed it
to timeout, where in fact it was a missing James Server, or a restarted name server.


Anyhow.  If I ever see the timeout again I will save the log!

-tim


On Jun 21, 2012, at 1:39 AM, Eric Charles wrote:

> On 06/20/2012 10:30 PM, Timothy Prepscius wrote:
>> I don't know yet.  Hasn't timed out again yet.
>> I hope it is not indeterminate.
>> 
>> I just tried to create a user, and it succeeded.  And it has been 8 hours since this
morning.  When I created a user.
>> Hmmm, I wonder what will happen if I restart the james server.
>> 
>> Ok, that succeeded too.
>> 
> 
> Strange. I would have bet on a broken connection.
> Eric
> 
>> 
>> Hmm.. I don't know.  I should have kept that log.
>> Will let you know the next time it happens,
>> 
>> -tim
>> 
>> 
>> On Jun 20, 2012, at 5:08 AM, Eric Charles wrote:
>> 
>>> Hi Tim,
>>> 
>>> An indication of the time needed to get the timeout will be useful.
>>> 
>>> We already have findbugs defines in the (project) pom, so if you invoke 'mvn
findbugs:check' you will get a findbugs.xml in the target folder. If you want something more
readable, you can invoke 'mvn site -Psite-reports' (look in site/findbugs.html).
>>> 
>>> Thx, Eric
>>> 
>>> On 06/19/2012 06:48 PM, Timothy Prepscius wrote:
>>>> K, I'll post a stack trace tomorrow.  (need to wait for the connection to
die)
>>>> 
>>>> 
>>>> [ optional ]
>>>> 
>>>> On a lesser note, do you guys do code coverage/static analysis of james?
>>>> 
>>>> Do you have a favorite tool/plugin that works well with maven and james?
>>>> (I'm not versed in the various maven java coverage/analysis tools)
>>>> 
>>>> I'll spend some time looking tonight, but if there is an obvious choice..
etc etc etc..
>>>> 
>>>> [ /optional ]
>>>> 
>>>> 
>>>> 
>>>> Thanks,
>>>> 
>>>> 
>>>> -tim
>>>> 
>>>> 
>>>> On Jun 19, 2012, at 10:23 AM, Eric Charles wrote:
>>>> 
>>>>> Hi Tim,
>>>>> 
>>>>> No idea on the overhead, the connect method instanciates a few objects
connecting to the remote server.
>>>>> 
>>>>> If your workload is not too high, it should be all right, but don't do
it via direct database access (never ever!).
>>>>> 
>>>>> It would be good that you post the exception stacktrace in a JIRA so
we can think to enhance the client with a JMX Connection pool (or something like that) or
configure the connection with longer timeouts.
>>>>> 
>>>>> Thx,
>>>>> Eric
>>>>> 
>>>>> On 06/19/2012 02:58 PM, Timothy Prepscius wrote:
>>>>>> I am wondering whether anyone knows the overhead associated with
the JmxServerProbe?
>>>>>> 
>>>>>> So, currently, I instantiate a JMX interface and keep it around indefinitely
in order to create users.
>>>>>> However, it eventually times out and fails to reconnect.
>>>>>> Thereafter, user creation fails.
>>>>>> 
>>>>>> 
>>>>>> So, if I use JMX, I will need to change the code to instantiate the
JMX interface with each user creation.
>>>>>> 
>>>>>> 
>>>>>> ---
>>>>>> 
>>>>>> Is this wise?  What sort of overhead am I inducing by creating this
JMX interface?
>>>>>> 
>>>>>> I'm wondering if I should instead just write directly to the james.JAMES_USER
table.
>>>>>> 
>>>>>> --
>>>>>> 
>>>>>> Thanks,
>>>>>> 
>>>>>> -tim
>>>>>> ---------------------------------------------------------------------
>>>>>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>>>>>> For additional commands, e-mail: server-user-help@james.apache.org
>>>>>> 
>>>>> 
>>>>> --
>>>>> eric | http://about.echarles.net | @echarles
>>>>> 
>>>>> ---------------------------------------------------------------------
>>>>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>>>>> For additional commands, e-mail: server-user-help@james.apache.org
>>>>> 
>>>> 
>>>> 
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>>>> For additional commands, e-mail: server-user-help@james.apache.org
>>>> 
>>> 
>>> --
>>> eric | http://about.echarles.net | @echarles
>>> 
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>>> For additional commands, e-mail: server-user-help@james.apache.org
>>> 
>> 
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
>> For additional commands, e-mail: server-user-help@james.apache.org
>> 
> 
> -- 
> eric | http://about.echarles.net | @echarles
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
> For additional commands, e-mail: server-user-help@james.apache.org
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org


Mime
View raw message