So, arguments about whether or not it's worth bothering aside (which is why I'm testing in the first place), I think I'm having some configuration issues in the below setup... A default ping test checks out ok to and from each to the others. Testing via ping -f -l 8972 A->B = fine B->A = fine A|B -> C = request timed out C -> A|B = request timed out I've trawled as much info online as I can find, and can't pinpoint the issue, if there is an issue. It's not the end of the world if I can't figure it out, everything works fine, and I can't complain about throughput at ~110MB/sec from any point to any other, but it's just bothering me. Any suggestions?
Try a smaller frame size. You're going to get TCP/IP and LACP overhead (are you trunking vlans as well?), which might be enough to cause your issue.
1472 seems to be the max, so clearly something isn't... I guess... working? Confession, when it comes to IP networking:
My next best guess would be something to do with the link aggregation. Could be the switch or NIC driver; no easy way to narrow it down though.
I'll continue poking things in the general vicinity, trial and error is always the best way. I just don't get why it times out rather than complain about needing to be fragmented
I wouldn't have thought using anything netgear for this kinda job would be your best bet. Looks like maybe the team or the switch can't do jumbo frames with lacp. Try breaking the team and cabling a single nic with jumbo frames for proof of concept perhaps?
It appears you may be right - I apparently completely neglected searching for issues with the switch, but it seems the 724T, LAGs and Jumbo frames are problematic - oddly most users are saying they're getting massive packet loss, but I'm not seeing any of that - in fact the only thing that suggests something isn't working is the ping test. The netgears represented the path of least resistance for STP support for Sonos (because it was doing some very odd things without it), so I'm not overly upset, fortunately. I'll give a switch independant team a punt and see how it get on. EDIT: FFS, found it. I was thinking the whole time that it would make sense if I need to set the MTU on both the ports and the LAG, but there's nothing under LAG configuration to do so. Because of course it's under port configuration, almost hiding in plain sight to click through to a page that's not accessible at all from LAG configuration. All sorted now, thanks for the pokes in the right direction gentlemen. I miss Brocade FC networks, where everything just makes sense (probably)
Just don't buy Dell switches! I have the best error with them right now. I have a stack of N3048s which won't run at 100Mb for 100Mb only devices. Just WON'T. Same port, same config, chuck in a gigabit device? .... Yeah works fine. Absolute rubbish!! I have a network colleague in southern US and he is chomping at the bit to take his N-series switch into the backyard with his 12-bore...
