8.2.3 - maybe a problem

[Paul A Vixie]

>   | I would have just converted the while to:
> 
> You mean the one I deleted?   What would that have achieved?
> Unles you're worried about needs continually being set, and so
> no events getting processed at all (which I think would be
> pretty unlikely)

that CAN happen.

> I'm not sure what good processing one event
> (if it exists) before processing the needs would do?

none.

> Is there some reason I'm not seeing why the needs (if there are
> some) can't just be processed first?  They should be rare (comparatively).

they can get into race conditions against event processing.  never fear,
i've figured out an architecturally clean way to do this, which will fix
the munnari problem without any help from douglas adams.

> In any case, as I said, munnari's nameserver has been behaving itself
> just fine since I made that change, and the one you suggested earlier.
> (Well, I have also disabled recursive queries, and I'm now getting 12MB
> an hour in syslog entries .. but that's an entirely different issue - the
> noise syslog makes about lame delegation, cnames in MX's, ... is all
> invisible amongst the "recursive query refused" messages)

there ought to be some way to route those refusals to /dev/null?