> I also would like to have more points of view on that. 
> Anybody here thinks
> that "give-up-on-error" behaviour can be considered a "feature"?

I think I've missed the context of the original post but here
is my 0.02 EURs:

In a mission critical environment where there are multiple smsc
connections kannel should always continue to work with other
connections despite one has failed. And try to make the
one that failed work again.

And more: when kannel crashes, dies, restarts, machine gets
restarted and kannel comes up and one of the smsc-s fails to
initialize, then kannel should not close up but continue
working.

I've had many bad moments with the last one. I have emi2, cimd
and smpp, can't remember right now tho which of them failed.
<sarcasm bitter="on">
Anyway you should outline in the kannel architecture doc that
kannel should be run one instance per smsc connection. That
should save people like me from trouble.
</sarcasm>

Regards,
Indrek


Reply via email to