* Ronciak, John ([email protected]) wrote:
> Hi Dave,
>
> Please see my comments in-line below.
>
> Cheers,
> John
>
>
> > -----Original Message-----
> > From: Dr. David Alan Gilbert [mailto:[email protected]]
> > Sent: Monday, February 03, 2014 12:11 PM
> > To: Ronciak, John
> > Cc: [email protected]
> > Subject: Re: [E1000-devel] dual e100 'exec cuc_dump_reset' vs PCI
> > latency (possibly vs Tulip)
> >
> > * Ronciak, John ([email protected]) wrote:
> > > Some list removed for now
> >
> > Hi John,
> > Thanks for the reply.
> >
> > > What do the HW stats for the failing port say?
> > > Is it receiving what it thinks are packets that a problem in some
> > way?
> >
> > I'm fairly sure they weren't incrementing at all, and I took a tcpdump
> > that was showing nothing coming from the e100's at that point.
> > Let me know which counters/debug to collect and I'll be happy to gather
> > it.
> Output the stats using 'ethtool -S <ethx>'. Do this before the failure and
> then again after. You can also get us the stack stats using 'netstat -s'.
OK, will do - might have to wait until weekend though.
> > > Is there a way to change the MTU being used on the camera that is
> > causing the problem?
> >
> > The camera doesn't give me much in the way of options (I'm reasonably
> > sure it's an embedded Linux of some form - but it doesn't let me at it)
> > If I change the MTU on my end of the link I guess it might follow?
>
> It should follow as it's negotiated during the connection set up. So please
> try that. It would be a good data point.
OK, will try.
>
> >
> > > Does moving the problem camera link to the other e100 port work?
> >
> > Note that the camera isn't on the e100 port - the camera is on the
> > Tulip; the PCs that are watching the cameras output are on the e100
> > ports, but it's the e100 ports that are stopping, and if I move the PCs
> > to another tulip port all works fine.
>
> So the 2 cameras are on 2 port of the tulip NIC and the 2 e100 ports are
> connected to other systems directly or to a hub or switch, correct? If you
> swap the two e100 ports does the problem follow the move or stay on that same
> port?
Not quite; I have:
switch (a) - has both cameras and one of the tulip ports
switch (b) - one e100 port and my desktop/laptop
switch (c) - other e100 port and someone elses desktops
Running the webbrowser on either b or c looking at the problematic camera,
but not the other, causes the e100 port to lockup (I think that's
only the e100 port that's viewing it - but again that's something I need
to check).
> > > Does the dmesg show anything else happening when the problem comes
> > up?
> >
> > Nope, nothing else.
> Ok, let's see what the stats show from above.
OK, will do.
Dave
--
-----Open up your eyes, open up your mind, open up your code -------
/ Dr. David Alan Gilbert | Running GNU/Linux | Happy \
\ gro.gilbert @ treblig.org | | In Hex /
\ _________________________|_____ http://www.treblig.org |_______/
------------------------------------------------------------------------------
Managing the Performance of Cloud-Based Applications
Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
Read the Whitepaper.
http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit
http://communities.intel.com/community/wired
