[storm] Storm and MySQL in WSGI environment - MySQL server run away

[james at jamesh.id.au]

On Wed, Jan 26, 2011 at 7:54 PM, Steve Kieu <msh.computing at gmail.com> wrote:
> As you can see the method call is already in try: except - I do not process
> if it failed yet but seems it just bomb out with 500 rather than going into
> the code res = e
>
> the meth (not listed here) will call a model object and model contains
> several Storm objects. I even try to reload the model in the __init__ of the
> Webnote class but it is still flaky, it work , sometimes it work some time
> got error but reload will work (improvement compared with before that it
> never recover until restart apache).

Where do your transactions begin and end?  In general, that is the
level at which you want to check for disconnections (or serialisation
failures and maybe integrity errors), since it represents the block of
work that will need to be redone.

If you are handling your transactions within the action_* methods,
then that is probably where you want to add your error recovery.

James.