Hello All,
This might be a tad long, so pelase bear with me.
We have been experiencing the same Bridged network problem as the comments in this post have suggested and yesterday we seemed to have solved this problem. Your mileage may vary, so this might not work for everyone, but I havent seen anyone cover this in the post, so it might be worth a go.
Just to clarify this is what was happening...
We have an IBM Blade Center (E series) running several HS20 and HS21 blades. We recently decided to trial Virtualisation and selected VMWare Server for this trial. We installed VMWare Server (1.0.5) onto our Blade HS20 running W2k3 and installed a W2k3 Guest and later a W2k Pro guest. Once we started trying to use the network we experienced problems. When first powering up a guest OS we would get brief network connectivity then after about two minutes the network connection would drop out.
We could ping between the guest and host, but not from the guest to the rest of our network or vice-versa.
We tried a number of solutions; all mentioend here (altering the registry on both the host and guest, updating drivers, trying all manner of stuff)
After much agro we determined that it might be best to setup another machine under Linux and try this. We had a new Blade just delivered and installed SUSE 10.3 as the host and installed the VMWare Server 1.0.5 again. We installed the same guest OS's and this time we had no problems what-so-ever. At this stage we purchase paid support from VmWare and logged out Windows Host problem with them.
After a day with no problem we believed the problem to be with Windows and after really liking SUSE and not hearing anythign from VMWare, inspite of logging the call as a Severity One, we took the decision to format the server and reinstall.
We installed SUSE 10.3 again and followed the same procedure to install VmWare Server 1.0.5 as before, this time, moving the working Guest installations to this new Guest server.
To our supprise the network problem reappeared. After about five minutes of head scratch we thought we had it figured.
The blades were all using Broadcom NetXtreme II cards, and at first I thought the new blade might have something different in it, but it was the same card. However what was different was the firmware version.
We checked our versios of the Blades we currently have against the latest downloads at IBM and found some new versions which seemed to match the new HS21.
After flashing the firmware of the Broadcom the problem appears to have been resolved. I've been runnign the server overnight and we have had 100% uptime throughout the night.
I'd though it's share this as I hadnt seen this mentioned and we saw that IBM was releasing a new firmware updater only a few days ago. I woudl encourage everyone to check their vendors download page for a new firmware revision of their Broadcom Network cards and indeed, while you're at it, the BIOS as well.
Hope this resolves the issue for everyone, again I fully expect that this wont solve everyones problem,but hopefully it might solve a few.
Regards,
Cai & Gav