We are having servers fail on Altiris multicast sessions. Ghost works multicast or unicast, Altiris unicast works (albeit more slowly on large deployments).
The network is built with Cisco 3750G switches, stacked, portfast and spanning tree on. The cables test fine on our Fluke tester. This is on our production network, so servers are moved and reimaged fairly often. We have also tried a Netgear unmanaged 10/100/1000 switch and a Cisco 3560, both with similar results; the 3560 imaged slower, but we had servers drop from the multicast the same.
Typically only about 1 in 10 servers fail, although we will usually have a single server fail out of smaller multicast groups. We are not having servers removed from the multicast because they are running slow. All fail with the error above.
Aside from dropping servers everything seems to be working well; the servers (including those dropped from the multicast session) report to the console, speeds are decent (about 3 minutes to push a ~1.25GB image), PXE and the Altiris boot process work flawlessly. Only multicasting is giving me fits.
Has anyone seen this before, or have suggestions?
Thanks.