Upgraded Network Infrastructure, Now Can't Image Computers

[scottpowers82]

Hello,

 

We recently upgraded our physical network infrastructure from Cisco Switches to Aruba S2500 switches. Now, this has upgraded us so that most of our network is now a full 1Gbps from the server to the workstation. However, I have noticed that when I try and image a machine (using WDS with MDT here...soon SCCM once I have some other problems figured out), the progress bar gets stuck at 5% and doesn't finish downloading the image. Also, I have noticed systems taking longer to PXE boot whereas it used to take maybe a few seconds to get the PXE response, it now seems to take at least 30 seconds or more. In fact, I have even had some systems NOT be able to receive the PXE response at all. I know it is probably something simple that we have to enable on the new Aruba switches, but don't know what setting it would be (plus, trying to learn Aruba's OS).

 

Anyone have any ideas?

 

Thanks in Advance,

 

Scott

[DMcCoy]

Some NICs have terrible PXE boot implementations, ADMtek, Nvidia especially.

 

Try changing the port speed at 100Mb as a test to see if WDS downloads the initial PXE boot quicker. Some machines I've imaged need dropping to 100Mb for boot and back to 1Gb one WDS starts to load for the image deployment.

 

If it's multicasting that the issue then check what the igmp settings are, slow DHCP is probably due to spanning tree, set clients to edge ports or use rtsp for the edge swiches.

[Sagima]

We found that some implementations of spanning tree broke pxe booting. It would take ages or just not happen.

[TronXP]

Do you use vlan's? The Aruba S2500's look very much like the Juniper EX3300's that I use.

[twin--turbo]

in ciscoland, seting ports as "portfast" stops them sending/recivind spanning tree BDU information. this means the port comes up and negotiates client links much faster.

 

We found PXE DHCP failed on switches without portfast.

 

worth investigating.