Re: [Patch] To add timeouts to SMPP connections

Hi Alex,

many thanks to you for soo good comparison...
But please do not mix 2 independend patches! One patch was for only add 
inactivity timeout and another one for sent queue cleanup.
Inactivity timeout patch is pretty simple and straightforward (imo).
Sent queue cleanup patch is more complex, but does the right thing and was 
extracted from centrium's tree, but nevertheless needs intensive testing 
phase before commit...

On Wednesday 20 August 2003 11:25, Alex Judd wrote:
> Never one to be precious about these things :)
>
> Alexander - thanks for the code which I've read through and basically
> there's a few pros and cons to the two approaches.
>
> 1. Aaron/Alex Patch
> - simple
> - Kannel-esque in style
> - not based around SMPP 3.4 protocol recommendations, but rather fixing a
> problem
> - does achieve the end result
>
> 2. Alexander's Patch
> - adheres to SMPP 3.4 protocol recommendations
> - fairly complex with changes to msg structures to add timing parameters
> and a new queue cleaning activity
> - however, really is the more accurate approach to the problem
> - potential duplicate sending problems with the requeue activity
> - default action (in current patch) is to requeue - would suggest as
> documentation says to change this to reconnect.

This behaviour was picked our emi driver, should we change it too ?

>
> So.. we have three basic choices.
>
> 1. Go the smsc_at / smsc_at2 approach and run the two in parallel (not
> recommended)

definitely -1 from me

> 2. Commit the complex patch straight away (potential risk to driver
> stability)
> 3. Settle with the simple patch
> 4. Get some volunteers to try the complex patch in the field for two weeks
> before commiting.

+1 from me , this was also the reason to post this patch to ML first ;)

>
> Personally I'd like to do 4., and I'll volunteer one of our connections to
> test it. Anyone else join me?

super, let me/us know how good these patches play for you... One thing I can 
say, we have these patches since ~3-4month on our production system running 
without any problems.

>
> Alex
> CTO
> Skywire
>
> ----- Original Message -----
> From: "Alan McNatty" <[email protected]>
> To: "Alexander Malysh" <[email protected]>
> Cc: "Alex Judd" <[email protected]>; <[email protected]>
> Sent: Monday, August 18, 2003 11:12 PM
> Subject: Re: [Patch] To add timeouts to SMPP connections
>
> > On Tue, 2003-08-19 at 09:45, Alexander Malysh wrote:
> > > sorry but Alex's patch is not closely conforms to spec. Please read
> > > smpp
>
> v3.4.
>
> > > issue 1.2 page 165.
> > > Quote:
> > > inactivity_timer - This timer specifies the maximum time lapse allowed
> > > between _transactions_ , after which period of inactivity, an SMPP
>
> entity may
>
> > > assume that the session is no longer active.
> > >
> > > Transactions doesn't mean enquire_link. I seen already some of SMSC's
>
> that
>
> > > doesn't undestand/reply to enquire_link. So this patch will just not
>
> work as
>
> > > expected. My patch doesn't depend on enquire_link and notice every
> > > transaction as proposed in smpp spec.
> > >
> > > Comments please...
> >
> > Ok - so it's more to do with any transaction rather than simply enquire
> > links .. my misinterpretation here - in this case I agree with you.
> >
> > Cheers,
> > Alan

-- 
Best regards / Mit besten Grüßen aus Düsseldorf

Dipl.-Ing.
Alexander Malysh
___________________________________

Centrium GmbH
Vogelsanger Weg 80
40470 Düsseldorf

Fon: +49 (0211) 74 84 51 80
Fax: +49 (0211) 277 49 109

email: a.malysh at centrium.de
web: http://www.centrium.de
msn: olek2002 at hotmail.com
icq: 98063111
___________________________________________

Please avoid sending me Word or PowerPoint attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html