I am running the PC version installed to a HD.  I have 3 3com 3c905
cards(bc's I think).  It is a P II 450 with a 10 GB IDE drive.  It has
over 3??MB of RAM. I was running 2-19-06 and 2-19-06 with the latest
update tarball applied.

Robert 

On Mon, 2006-02-27 at 13:39 -0600, Bill Marquette wrote:
> On 2/27/06, Robert Goley <[EMAIL PROTECTED]> wrote:
> > I will retest with Beta2.  I had the same results that John reported
> > with Vonage lines.  I only had to test it with one of the lines.
> >
> > Robert
> 
> Thanks...the workarounds kinda suck IMO and we're still seeing issues
> on WRAPs (but not all of them!?!?!?!) so I won't be entirely surprised
> if you come back and say it still sucks ass.  This appears to be a
> FreeBSD issue, on same hardware (Soekris 4801) on OpenBSD, this works
> like it's supposed to.
> 
> I've spent at least 12 hours this last weekend tracking this crap
> down.  The source code between FreeBSD and OpenBSD for ALTQ is line by
> line nearly identical, the major differences have to do with locking
> differences and minor other porting items.
> 
> Here's what we've learned so far:
> 2/15/06 snapshot w/ upperlimit set sucks on WRAP and 4801 (sis driver)
> 2/15/06 snapshot w/ upperlimit set works fine on Nexcom 1030 w/ Intel
> nics (fxp driver)
> 2/15/06 snapshot w/o upperlimit set works fine on 4801 (sis driver)
> 2/15/06 snapshot w/o upperlimit set works fine on Nexcom 1030 w/ Intel
> nics (fxp driver)
> 
> Beta2 w/o upperlimit set works fine on 4801 (sis driver)
> Beta2 w/o upperlimit set sucks on 2 of 3 WRAPs we have (sis driver)
> Beta2 w/o upperlimit set works fine on Nexcom 1030 w/ Intel nics (fxp driver)
> 
> OpenBSD 3.9-beta works fine w/ upperlimit set on 4801 (sis driver)
> OpenBSD 3.9-beta works fine w/o upperlimit set on 4801 (sis driver)
> 
> The removal of upperlimit sucks because that means we can't limit the
> bandwidth any queue is allowed.  We're seeing evidence of a 10x
> multiplier bug, but upperlimit isn't exposed to the NICs - 960Kbit
> upperlimit gives me great Vonage calls (92 or 96Kbit codec, can't
> recall).   128Kbit upperlimit gives crappy calls as does 512Kbit or
> 768Kbit.  To me this is pointing to a multiplier broken somewhere, but
> I haven't yet found it and it _appears_ to mainly affect the sis
> driver (although I wouldn't be surprised in the least if it didn't
> affect others).
> 
> What would be most helpful right now is a report of people running b1
> snapshots and have tried the shaper.  What we need is the hardware
> you're running on:
> CPU
> NIC(s)
> Install type (embedded, pc, iso, etc)
> Whether it works or sucks ass
> 
> I'll resend this request on the blog and as a seperate email later
> today.  Thanks
> 
> --Bill
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to