XYMer's Home away from Home

Sorry for my bad mood, but I've been trying to get this server working for a week, working 20+ hours per day on it, and nothing wants to work together. This is more of a rant than a request for help.

<rant>First, I had a problem with the second Ethernet port not working. Apparently Intel wanted to put the latest and (not) greatest chip for the second controller. After some research, I found that this controller requires a linux kernel equal to or newer than 2.6.31. Debian's latest stable release, Lenny, which I had been using, uses 2.6.28. So, I thought I'd try my hand at compiling a new kernel. It successfully compiled, and I installed it. However, on boot, it would hang. I'm not sure exactly why. It didn't really say anything about it.

I booted back to 2.6.28 and then tried updating the apt-get config file to search "testing" releases instead of just stable. In the process of updating, something messed up, and hosed the system. So, I formatted the drive and installed Debian Lenny again.

I did more research and found that supposedly the 82578DM supposedly only requires the e1000e driver, which is compiled into the kernel. I tried downloading the latest e1000e driver and compiling it. After unloading and reloading e1000e, the second NIC showed up, but wouldn't get an IP address. I thought maybe restarting would fix, but after that, the second NIC no longer showed up. I could check the version of e1000e, and it showed that it was the new one, but I noticed the size was the same as the old version (I can't remember the commands I was running - this was a few days ago). After unloading and reloading e1000e, the second NIC showed up, but still had no IP. Even setting it manually wouldn't save it.

So, I decided to download the Debian Squeeze ISO, even though it's still in testing. It's using a 2.6.32 kernel, which I hoped would work with my second NIC. I installed it, and it booted fine. I checked and saw that both NICs were recognized right after booting, and both had IPs.

Happy that everything seemed to finally be working, I installed VMWare Server 2.0.2. I had installed it back when only one NIC was working, but that install was lost when the earlier attempt to update to testing failed. This time I had to find a patch to the setup script in order to install it properly, but it seemed to work, and I was able to get several VMs set up and running.

However, this morning, I noticed that the web UI said that the plugin had lost its connection to the server. I restarted the vmware-mgmt service, but still couldn't connect to the server. I tried restarting vmware, but it said an instance of vmware was running, and couldn't be restarted until it was stopped (why have the restart option if it can't restart?). I tried stopping it, but got another error. I did some more searching, and apparently VMWare Server doesn't work properly in 2.6.31 or later kernels (remember that just happens to be the minimum kernel for my second NIC to work). I restarted the computer, hoping that would get it working again, and hoping that it was something I wouldn't need to do very often, as this will be a production server.

That brings me to where I am now. The Grub rescue screen. I have one OS installed on the server (unless you count VMs, which don't count). I have changed nothing in terms of hard drives or Grub since my previous boot, which went fine. This isn't the first time I've been dumped to the Grub rescue screen either, after having done nothing that should have affected Grub. I might be able to get the computer to boot again, but it shouldn't be here in the first place. I can't have a production server that needs to be restarted frequently because VMWare Server can't work properly, then boots to a Grub rescue screen all the time.</rant>

When i get into loops like that i get up take a walk and get an ice cream
GL dont worry you"ll get it.

What kernel is Ubuntu Server on now?

Have you given any thought to running RHEL, which may/should have different kernel requirements than a Debian-based distro. If Intel doesn't support RHEL they're going to be cut out of a lot of enterprise sites, I doubt they'd destroy their market share. If that works, you could try switching to CentOS which is RHEL without the required support subscription (well, not required, but the official update system doesn't work w/o a subscription).

I approach building linux machines just like I do hackintoshes. Try to make sure every single component is supported, otherwise it's out of the cart. I hate chasing down each and every little driver or conflict. Good luck to you!

Yeah. I had previously already decided on a different motherboard by Asus, which has two of the supported NICs, but thought Intel would be the safer bet, especially since it had more PCIe slots in case I needed to add a second RAID card or 3rd/4th NIC. Also, the $30 off combo deal with the Xeon ended on the Asus and went to the Intel board. I should have stuck with what I'd originally decided on (the Intel board isn't bad. It's just that the second NIC is too new).

New annoyance: Since the main reason I had switched from ESXi to running VMWare Server in a Linux host OS was because of the second NIC, and that wasn't working out so well, I decided to give ESXi another shot. It is nice to not have to have a host OS, and ESXi allows me to allocate the CPU cores more like I wanted (I can allow a guest OS to use all 8 virtual cores rather than only 2). The vSphere Client also works better than the VI web interface, which was required for Server. For some reason though, even though ESXi only takes a few seconds to boot, it takes 11 minutes before it's usable. The host will say I can log in and manage it at the displayed IP, but it refuses connection. Guest OSes also don't start during this time. I can however ping the host and SSH into it just fine. It boots from an old USB stick, so my first thought was that there was something loading slowly from that, but the activity light isn't flashing. I'm wondering if VMWare added in a delay so people would pay for vSphere instead of using the free ESXi.

Hopefully it won't be that big a deal. I don't plan on rebooting the host very often (preferably never, but I'm sure times will come up when it is required).