Kent IV, William (WW) wrote:
I've got an almost identical problem, except I'm using a 3Com 3924 GigE switch (and Dlink DGE-500T adapters). Also, the motherboards have on-board 10/100 connections that aren't being used.
Yeah, that sounds like a horse of an entirely different color. Maybe the gig-E driver for whatever chipset's in the D-Link cards was "improved" during this time period. Some kernel-fu could sort you out in that case, or you can check the kernel.org changelogs to see if any modifications were made in the interim.
Mmmm, kernel-fu ...
I've always thought it was a problem with the Dlink adapters and/or their drivers. Maybe I need to pay for some support from RedHat? I'll try Steven's suggestions as well, but probably not until after the holidays.
Well, the line of reasoning for the mcast_ttl value change is, "Hey, maybe the network stack is handling TTL values differently for multicast from one kernel version to another."
Come to think of it, I did have to monkey with the mcast_ttl values a bit when I first deployed the monitoring core on the production boxes, and that was a 2.4.1x kernel at the time...
Ho ho ho...
