Finally found a way to image 20 PCs at once without a server hiccup

Been fighting with multicast imaging for ages, every time I tried to push a Windows image to a full lab the switch would choke and half the machines would drop off. Tried adjusting the IGMP snooping settings, even swapped to a managed switch, still got random failures. Then I stumbled onto setting a fixed multicast range on the DHCP scope, stopped the broadcast storm before it happened. Last deployment at a community college here in Austin, all 20 units finished in 12 minutes flat, not one dropout. Anybody else have weird luck with multicast that turned out to be a DHCP config issue?

2 comments

2 Comments

spencer_hayes7129d ago

Actually the DHCP thing is more about stopping the machines from fighting over IP addresses when they all boot at once. The multicast range setting helps keep the traffic from flooding the network, but it doesn't really stop the broadcast storm you mentioned. That storm usually comes from the switch not handling the multicast groups right, even with IGMP snooping on. So you might have fixed two separate problems without knowing it. The fixed DHCP range probably just made each PC get its address faster and cleaner, while the real multicast fix was something else you changed along the way. Either way, 20 machines in 12 minutes is solid. Just don't expect that same DHCP trick to fix every lab issue, cause it's only half the story.

amy9751mo ago

Saw a tip once that said multicast issues usually hide in DHCP, glad yours was an easy fix.