I have a new Dell PE R810 that we have deployed ESX on. It is version 5.0. The reason I am using 5.0 is because there was not a hard lock on the maximum physical memory.
Server specs...
4 x 2.0GHz 8c e7-4820 XEON
256 GB of RAM.
4 x 500Gb HDD (Raid 5) using PERC
Two Dual-Port Broadcom® 5709c gigabit NICs integrated on Motherboard.
2 port 10G board ( I can't remember the specs on this one, but it is totally unused)
2 single port intel network cards (added in case they were needed for troubleshooting)
My problem is that about two times a week, or so (randomly), I start losing network connectivity to the server. The switch doesn't seem to show anything, but the event log on the ESX server starts to show that it vmnic0, and vSwitch 0 are down, and then back up. The message is "Lost network connectivity on virtual switch "vSwitch0", Physical NIC vmnic 0 is down." and then a random time later (seconds to minutes) I get the message telling me it is back up. This will continue usually until the server is rebooted. I cannot see the logs for the network switches, because they are handled by our networking team. I can see a summary of the switch port, and how many collision, dropped packets, etc..., but nothing looks bad there.
I thought that it might be the 1G network port it was originally plugged into, off the Cisco switches we have here, so I switched it to a 100Mb, connection Cisco as well, and the problem returned. I am going to try and move it to one of the Intel cards I installed in the server later today, to see if that works better. If it does, then I can assume that either my broadcom network cards have a problem. In which case I will have to have Dell take care of it.
This is currently used for a Pilot course where there are only 15 VM running, 12 - Ubuntu, 2 PFsense, and 1 - MS Server 2003. So the server is totally under worked.
Any ideas on test or information, that can help me diagnose this problem? Are there any know issues with the broadcom cards? I can post logs is somone asks.
Any help would be greatly appreciated.
Thanks in advance
Jeremy
