Guys I have a VMware virtaul server that is using enhanced NIC so has jumbo frames and tcp offloading enabled. And we are seeing performance issues to the app on this server. When we capture the traffic with wireshark on the server we see a huge amount jumbo frames/packets (anywhere from 2500 to 32000 bytes), that we are not seeing on the network or end client, as well as of a lot tcp errors, in terms of bad checksums and retransmissions. On top of this each of the jumbo frames/packets have the DF bit set, which immediately tells me that these packets will be dropped by the lower default MTU on the network devices, due to no fragmentatoin occurring.
My problem is Im struggling to pinpoint this and find where these packets are being dropped if at all. I was hoping to see this maybe from the interfaces counters on the switches but I see nothing. Or maybe this is happening at the physically interface of the VM host server? Is my logic around this correct or am i missing something? Cheers Stu -- Stuart Hare [email protected]
