On Sat, Dec 04, 1999 at 09:52:07PM +0100, Ignatios Souvatzis wrote:
> On Sat, Dec 04, 1999 at 12:13:19PM -0500, Bill Sommerfeld wrote:
> > we should probably arrange for the lcp keepalives to go into the
> > lowdelay queue..
> 
> Uhm...
> 
> i'd say this is a broken worksaround for a broken workaround for OSes that
> dont have reliable SIGHUP signalling (namely, it was introduced for LInux).
> 
> LCP echo is a diagnosing tool... using it in dumb mode as HUP detection
> replacement is wrong in the first place. Said timeout should be stopped by
> any received (valid) packet, not just by LCP echoreply

Yes, I think thiswould be a good solution.

Why am I using LCP in the first place?  Because for some (unknown) reason,
the PPP link tends to lock up (sometimes it takes several weeks before this
happens, sometimes only hours or so) and the simplest way to automatically
recover from this is to use LCP echo.

  -jm