Comments on: Link State Tracking in Blade Deployments http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/ The weblog of an IT pro specializing in virtualization, storage, and servers Wed, 08 Feb 2012 17:13:47 +0000 hourly 1 http://wordpress.org/?v= By: A. Mikkelsen http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-49262 A. Mikkelsen Wed, 29 Sep 2010 19:20:34 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-49262 Thanks for this solution. It helped us. A. Mikkelsen Thanks for this solution.
It helped us.

A. Mikkelsen

]]> By: William http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-47591 William Thu, 25 Feb 2010 04:31:39 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-47591 Very nice, thanks for the info, needed to do some work tonight and this explains a line which I didn't fully understand 5 minutes ago :) Very nice, thanks for the info, needed to do some work tonight and this explains a line which I didn’t fully understand 5 minutes ago :)

]]> By: Chris http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-43665 Chris Thu, 19 Feb 2009 17:05:54 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-43665 Rather a late chime in, but I just experienced major problems with 'Beacon Probing'. I had four GB NICs setup in a channel group as outlined in your article. However I was getting reports of very poor performance. Using iperf I was able to determine that typical network speed through this was ~50Mbps. If I then put the same server in a portgroup with a single NIC it could run up to 700Mbps. I tried plenty of things before eventually disabling Beacon Probing at which point the network performance returned to the expected 600-700Mbps. I don't claim to understand what was happening, but I can do without beacon probing. Rather a late chime in, but I just experienced major problems with ‘Beacon Probing’. I had four GB NICs setup in a channel group as outlined in your article. However I was getting reports of very poor performance.

Using iperf I was able to determine that typical network speed through this was ~50Mbps. If I then put the same server in a portgroup with a single NIC it could run up to 700Mbps.

I tried plenty of things before eventually disabling Beacon Probing at which point the network performance returned to the expected 600-700Mbps.

I don’t claim to understand what was happening, but I can do without beacon probing.

]]> By: slowe http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-41899 slowe Fri, 10 Oct 2008 12:53:00 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-41899 James, thanks for the clarification. I've modified the post accordingly to help avoid confusion. James, thanks for the clarification. I’ve modified the post accordingly to help avoid confusion.

]]> By: James Cape http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-41897 James Cape Thu, 09 Oct 2008 18:59:41 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-41897 Terminology quibble: The term "Chassis Switch" refers to something else: a modular switch, e.g. the Cisco 6500 series. Terminology quibble:

The term “Chassis Switch” refers to something else: a modular switch, e.g. the Cisco 6500 series.

]]> By: slowe http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-40343 slowe Wed, 06 Aug 2008 13:24:07 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-40343 Mark, Interesting issue. Send me an e-mail and let's take this discussion private. I'll be happy to help, if I'm able to do so. Mark,

Interesting issue. Send me an e-mail and let’s take this discussion private. I’ll be happy to help, if I’m able to do so.

]]> By: Mark Burse http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-40338 Mark Burse Wed, 06 Aug 2008 08:11:10 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-40338 Hi Scott, I have an intersting one for you:- I have a C7000 with 6 x VC switches - 4 x Ethernet and 2 x FC. Yes I am using the Ethernet Switch Link cables. I can get the 2 x vertical ones working on the left-hand-side working, but cannot get the other 2 x switches to play together nicely….everything points to the customer’s 2nd core switch as the issue… Strangely when I use just the left-hand-side switches I only get 1 x active connection…which is great…ALL Works no issues – But when I enable the 2 x right-hand-side switches as well, it drops my active connection from the left-hand-side and then activates 2 x Active connections causing a Loop so drops all packets…..I believe? All Hardware is running latest F/W as of today. We are using A "Shared Uplink Set"....Server are ESX 3.5.0 Update 2. Rgds Mark Hi Scott,

I have an intersting one for you:-
I have a C7000 with 6 x VC switches – 4 x Ethernet and 2 x FC. Yes I am using the Ethernet Switch Link cables.
I can get the 2 x vertical ones working on the left-hand-side working, but cannot get the other 2 x switches to play together nicely….everything points to the customer’s 2nd core switch as the issue…
Strangely when I use just the left-hand-side switches I only get 1 x active connection…which is great…ALL Works no issues – But when I enable the 2 x right-hand-side switches as well, it drops my active connection from the left-hand-side and then activates 2 x Active connections causing a Loop so drops all packets…..I believe?
All Hardware is running latest F/W as of today.
We are using A “Shared Uplink Set”….Server are ESX 3.5.0 Update 2.
Rgds
Mark

]]> By: slowe http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-35080 slowe Tue, 15 Jan 2008 21:42:31 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-35080 Erik, I'm glad you found the information helpful. Feel free to spread the word about the site! Erik,

I’m glad you found the information helpful. Feel free to spread the word about the site!

]]> By: Erik http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-35073 Erik Tue, 15 Jan 2008 18:16:41 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-35073 Thanks for this great information. It came in very handy on a new BladeCenter H implementation with ESX 3.5 Thanks for this great information. It came in very handy on a new BladeCenter H implementation with ESX 3.5

]]> By: slowe http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/comment-page-1/#comment-34820 slowe Sat, 29 Dec 2007 04:01:12 +0000 http://blog.scottlowe.org/2007/06/22/link-state-tracking-in-blade-deployments/#comment-34820 Sonia, I, too, have heard of problems when using beacon probing, but I don't have any direct experience one way or the other. Hopefully some other readers will be able to share their experience and/or knowledge on the use of beacon probing. Anyone care to chime in? Sonia,

I, too, have heard of problems when using beacon probing, but I don’t have any direct experience one way or the other. Hopefully some other readers will be able to share their experience and/or knowledge on the use of beacon probing.

Anyone care to chime in?

]]>