ws-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Petersen, Jennifer" <Jennifer.Peter...@cbsa-asfc.gc.ca>
Subject RE: Logging to SystemErr not acceptable.
Date Wed, 22 Feb 2006 15:08:04 GMT
Thanks John.
 
For sure I am trying to fit this into our environment, which for now will not change. The
logging configuration and monitoring is out of my hands.
But not upset at all. Actually thrilled anyone out there replied at all.
And I don't "expect" anyone to do any work .. I can do the work but appreciate any help I
can get.
 
My assumption was that Java supported a "synchronous" timeout at the socket layer and so this
should be a simple feature to add.
I will have to look at this more closely myself. - I might be able to get away with a TimerTask
in the end.
 
thanks!
 
----Original Message-----
From: John Buren Southerland [mailto:john@southerland-consulting.com]
Sent: February 21, 2006 5:33 PM
To: xmlrpc-user@ws.apache.org
Subject: RE: Logging to SystemErr not acceptable.



Jennifer,
I am as eager to help add to this system as anyone else, but please try not to force it to
exist within your environment irrespective of how others are using it.  Perhaps the issue
here is to find a better monitoring program that can exclude certain messages from being paged
to the people who are supporting it, in order to save them the constant joy of seeking employment
elseware. (Perhaps in a shop where turnover is not initiated upon server restart)
I know at times it can be frustrating to get a solution to fit within a ridgid corporate policy/framework,
but that is hardly the burdon on everyone using the library.  Perhaps you should look deeper
into configuring the log4j logger to do what you need with messages you choose not to see.
I'm really not trying to be devisive, but you do seem pretty upset with something that sounds
configurable to me.  
On another topic I will see if I can manhandle my user timeout solution into the httpclient
library and see if that core team likes it or not.  I believe the httpclient object can be
accessed from a method within XMLRPC so that timeout should apply here as well.  Unfortunately
the 1.5 dependence tells me it may wait in the wings.  I doubt even that your websphere app
server is running anything newer than 1.4, so that may not even help you.  In fact if it is
running 1.3 socket timeout probably isn't even supported.
Again remember this is opensource, so, ehh, maybe next week.....
Good Luck, John

On Tue, 2006-02-21 at 16:57 -0500, Petersen, Jennifer wrote: 

I don't think the administrators of the application server would appreciate it if I redirected
the standard error output stream of their Websphere instances. :)



Seriously, in the real world here. Our application is hosted on a System running Websphere.
Many applications run there. The administrators decide on the deployement configuration and
how logging is to be done. The administrators of that System monitor the logs vary closely.
The system is HA. People get paged in the middle of the night when errors get logged. People
get fired when system goes down.



I stand by what I said. Logging like this is inappropriate in library code. 



anyways. its a deal breaker for me. but just trying to convince anyone who will listen why
 why you shouldn't be doing this.



-

Jennifer



-----Original Message-----

From: Michael Landon - IBN [mailto: mlandon@ibnads.com]

Sent: February 21, 2006 2:56 PM

To:  xmlrpc-user@ws.apache.org

Subject: Re: Logging to SystemErr not acceptable.





http://www.devx.com/tips/Tip/5616





----- Original Message ----- 

From: "Petersen, Jennifer" < Jennifer.Petersen@cbsa-asfc.gc.ca>

To: < xmlrpc-user@ws.apache.org>

Sent: Tuesday, February 21, 2006 11:57 AM

Subject: RE: Logging to SystemErr not acceptable.





I can't. I do not have control over the environment.



I would argue that it is incorrect for library/infrastructure code to be 

logging this way.

I need to control how/where/when errors and messages are reported. In this 

case it is a critical app where logs are closely monitored. Even so, I can't 

see this requirement being terribly unusual even for non-critical apps.



In any case I need to get rid of this. or not use the library :(

So is the approach to create my own patch?



thanks.



-

Jennifer





-----Original Message-----

From: Schölver, Andreas [mailto: Andreas.Schoelver@ebootis.de]

Sent: February 21, 2006 12:39 PM

To:  xmlrpc-user@ws.apache.org

Subject: RE: Logging to SystemErr not acceptable.





Simply redirect shell output on System.err to /dev/null.



Andreas



> -----Original Message-----

> From: Petersen, Jennifer [mailto: Jennifer.Petersen@cbsa-asfc.gc.ca]

> Sent: Monday, February 20, 2006 4:25 PM

> To:  xmlrpc-user@ws.apache.org

> Subject: Logging to SystemErr not acceptable.

>

>

> I need to stop the client from logging to SystemErr.

> Any help would be appreciated.

>

> Here's an example:

>

> [20/02/06 9:44:10:015 EST] 60936093 SystemErr     R Avoiding

> obscuring previous error by supressing error encountered

> while ending request:

> org.apache.xmlrpc.XmlRpcClientException: Exception closing

> URLConnection

>

> --

> Jennifer Petersen

>







Mime
View raw message