1. We need to handle network related errors at navigation level and make
   sure that we capture them and translate to appropriate business errors and
   then a mechanism should take it over from there.
   2. Whenever the server stops responding in between, or there is network
   breakdown, these issues crop up.
   3. We can write a wrapper code to validate whether the navigation really
   happened or not by having a framework for applying regex'es and validate.
   Hence for every navigation, we should have positive expects and negative
   expects. +ve expects tell us that navigation was success when the response
   has there strings. -ve expects help us identify that the navigation was not
   a success.


Yours sincerely,

Sridhar Vedhanabatla


On Mon, May 30, 2011 at 10:40 AM, Manikandan R <[email protected]> wrote:

> Hi Everyone,
>
> I've started using apache commons httpclient library (version no. 3.1) to
> make the http calls recently. In production environment, seeing the
> following exceptions in logs:
>
> 1. Broken Pipe
> 2. Connection Reset
> 3. Connection Timeout
> 4. Read Time out
> etc
>
> Whenever these exceptions, we get the complaints from customers not
> completing the tasks. So, we repeat the work manually. To avoid doing it
> manually, program should take care of the retries without causing any
> duplicates. I am not sure when to do the retry. Basically, for what type of
> exceptions, Should I do retry so that it won't end up in duplicates?
>
> Please suggest.
>
> Thanks,
> Mani
>

Reply via email to