Cannot resolve peculiar, persistent NIC error in WHS 2011 Event Logs

[BonzaiDuck]

So . . . In the process of some tidy-up troubleshooting on my super-fast workstation, I made great progress in tracing down event-log red-bang entries to their source and resolving them. Once the workstation was "all blue," I decided to pursue the same initiative on my WHS 2011 server.

Now . . . to describe that server. [And yes -- I know -- WHS is "orphaned." I should spend $300+ on "Server Essentials 2000-whatever." I'm resolved to stick to what I have.]

This was a seemingly convenient use of spare parts, but I didn't think about the orphaned drivers. I resurrected an old ASUS Striker Extreme 680i mobo by simply replacing the BIOS chip. I had a lot of DDR2 memory sticks, and a barely-used Conroe E6600 C2D. So I installed the WHS 2011 (Server 2008 R2) OS.

Then! I struggled to get the necessary drivers. Very troublesome, but it turned out that the Win7-64 drivers for this chipset worked fine with WHS.

This server has been working fine for at least 2+ years -- 24/7 nonstop. It has the StableBit add-in, the Adv Admin Console add-in, and the Lights-Out add-in -- which pretty much manages the sleep-states and backup schedule for four other workstations in the household. I just never thought to focus on the Event Logs with a "fine-toothed comb" -- so to speak.

All but ONE of the red-bangs has been purged. Here's the summary description:

Log Name: System
Source: NVNET
Event ID : 5008
Level: Error
User: N/A
General Tab description:
NVIDIA nForce 10/100/1000 Mbps Ethernet : Has encountered an invalid network address.

System Behavior: Booting up and logging in to the WHS desktop, it seems to take forever before the network connection(s) are established. The little "TV-set" icon is busy with the blue circle for what seems like a few minutes, before it reports connection with "internet access." Otherwise, once the bootup is complete (and without ever logging in or accessing the server desktop) -- it is fully functional.

Actions Taken:

1) Uninstalled the nForce drivers and then reinstalled them -- E-ID 5008 persists . .
2) Changed automatic assignment of the server's IP address to a fixed IP outside the range of the DHCP assignment but within the subnet. Specified the household router as gateway. Entered the three DNS IP addresses for my ISP's servers. E-ID 5008 persists.

3) (and especially) resolved all other red-bangs, which were a matter of granting security logon access to "NT Authority . . ." so that initializing services don't hang and need to "retry."

4) One of the red-bangs was for nForce IAM -- a service intended to interact with the desktop, but for security reasons, WHS doesn't allow that, and raises the error. The solution was to edit the service's properties and uncheck the "interact with desktop" box.

Results: Now it doesn't seem to take forever for WHS to report full internet access and all the network handshaking success. Which is good . . .

But . . . Event ID 5008 continues to appear at bootup, and here or there as the server continues running without mishap for anything else -- including video streaming and workstation backups.

I thought I'd post this here -- somebody may know something I missed.

The next step will be disabling the onboard NVidia network controller after physically installing either a Trendnet TEG-PCITXR PCI G-bit card, or an Intel "Pro" PCI_E x1 card -- which arrives next Monday. A little poking around on the web indicated that WHS 2011 has built-in drivers for the TEG's Realtek chip, and it also has them for the Intel Pro NIC I ordered. [Might have avoided ordering the Pro, but I was a day late in discovering the TEG card in my parts locker.]

 

[Geofram]

The next step will be disabling the onboard NVidia network controller after physically installing either a Trendnet TEG-PCITXR PCI G-bit card, or an Intel "Pro" PCI_E x1 card -- which arrives next Monday. A little poking around on the web indicated that WHS 2011 has built-in drivers for the TEG's Realtek chip, and it also has them for the Intel Pro NIC I ordered. [Might have avoided ordering the Pro, but I was a day late in discovering the TEG card in my parts locker.]

Sounds like you're already doing what my next step would be.

I'm not sure what error that event refers to, but throwing a different NIC in the computer and disabling the nVidia one should at least tell you if the error is configuration related or hardware (or driver) related. My gut says that it's something with the nVidia driver.

 

[BonzaiDuck]

Sounds like you're already doing what my next step would be.

I'm not sure what error that event refers to, but throwing a different NIC in the computer and disabling the nVidia one should at least tell you if the error is configuration related or hardware (or driver) related. My gut says that it's something with the nVidia driver.

Well, the lights on my ol' Netgear g-bit switch tell me that the NVidia nForce controller is negotiating 1,000 Mb in full-duplex. The menu for IAM before and after I disabled the IAM service's "interaction with desktop" only shows 10 HD/FD, 100 HD/FD and "Auto Negotiation" or whatever it's called. The "Network Connection" window's properties tell me that it is indeed a gigabit connection, as do the volume of either incoming or outgoing would suffice to do.

Now the latest development: I decided to run Windows Update from WHS'11. I had it go through and rediscover any needed updates.

Thing comes back with a Windows update based on a KB article, and it was a set of replacement files for network RPC1 that "resolved problems" for WHS networking and made for both better connection reliability and speed.

So now, the question is whether NVidia gb Ethernet controller throws up that same error at boot up next time, or goes away. Before the RPC update, the NVNET 5008 error would appear in the logs every several hours over a day's time between reboots. It's not doing that now.

If I have to do a little work with the machine apart in this cramped room, I'm gonna do it. But I can take my damn time now, too.

 

[azazel1024]

Yeah, I'd pretty much do what you are doing. Disable the onboard NIC and slap in a new one. Have you been using a static IP in the machine? Also where is the error popping up? System events? Or DHCP Client events? Looks like system, which sounds like you have a static IP set for the machine.

Based on a 5008, stab in the dark is that there is a static IP assigned to it and another machine on your network has the same IP address, possibly handed out by your router. Ensure that the static IP of the server is NOT within the dynamic IP address range of the router DHCP server.

The other thing is...I am a big believer or reusing hardware when possible, but if this is a 24/7 server, odds are good you are probably tearing through a lot more electricity than if you just put together a very cheap, low power machine.

My Ivy Bridge celeron based G1610 uses around 21w at idle and 33w streaming. About 50w under max load. Whole build, excluding drives, was about $250 (CPU, case, 380w bronze rated PSU, MoBo, RAM).

I have no earthly idea how much your machine is running, but is it is similar to my old Core 2 duo E7500 based machine that I got rid of a couple of years ago for an i5-3570 based machine...its probably 60-80w at idle...which depending on your electric rates, is probably $30-100 more per year than a nice low power server would use.

You want even cheaper, get one of the mITX Bay Trail based celeron boards that came out recently...for around $60-70 you have the complete system, other than RAM, case and PSU (and drives)...but I'd guess you have all the rest you need, other than the RAM. $100 could have you a decent home server saving you possibly that in electricity in just a year or year and a half (since the bay trail based board probably uses a further 5-10w less than what my machine uses).

 

[BonzaiDuck]

Yeah, I'd pretty much do what you are doing. Disable the onboard NIC and slap in a new one. Have you been using a static IP in the machine? Also where is the error popping up? System events? Or DHCP Client events? Looks like system, which sounds like you have a static IP set for the machine.

Based on a 5008, stab in the dark is that there is a static IP assigned to it and another machine on your network has the same IP address, possibly handed out by your router. Ensure that the static IP of the server is NOT within the dynamic IP address range of the router DHCP server.

The other thing is...I am a big believer or reusing hardware when possible, but if this is a 24/7 server, odds are good you are probably tearing through a lot more electricity than if you just put together a very cheap, low power machine.

My Ivy Bridge celeron based G1610 uses around 21w at idle and 33w streaming. About 50w under max load. Whole build, excluding drives, was about $250 (CPU, case, 380w bronze rated PSU, MoBo, RAM).

I have no earthly idea how much your machine is running, but is it is similar to my old Core 2 duo E7500 based machine that I got rid of a couple of years ago for an i5-3570 based machine...its probably 60-80w at idle...which depending on your electric rates, is probably $30-100 more per year than a nice low power server would use.

You want even cheaper, get one of the mITX Bay Trail based celeron boards that came out recently...for around $60-70 you have the complete system, other than RAM, case and PSU (and drives)...but I'd guess you have all the rest you need, other than the RAM. $100 could have you a decent home server saving you possibly that in electricity in just a year or year and a half (since the bay trail based board probably uses a further 5-10w less than what my machine uses).

All your advice is generally practical, but not so feasible for me at the moment -- although $30 on the annual power-bill might eventually be worth the increment of trouble for the increment in savings. But "flows" of savings require a "stock" of parts investment, or the whole thing boiled down to "net present values." I might just break even for keeping the old rig for a few more years.

The server box was getting an IP assignment from the router's DHCP server in the specified range for the household. I DID change to a fixed IP outside that range but in the subnet, and the message continues to occur at boot time, even after the Windows Update I mentioned.

In all other respects, the NVidia net-controller works without flaw, and I went from two related errors to one during the course of my cleanup.

But I was just handed the package that contains the new Intel Pro NIC and a 60GB SSD I plan as replacement to my current boot-system drive for WHS. I'll break even on the power consumption, since I just added another 1TB drive to my pool, and the current boot drive draws the same amount of power.

If I discovered that I couldn't get all the drivers for this NVidia 680i board compatible with Win 2008 R2 (and WHS2011), I'd be ordering new parts or reconsidering my strategy with NAS options.

I can't help but encourage comments about what I intend to do today. I was networking my PCs as early as 1993 -- when we used NetBeui and those coax connections. Here and there, I'd always spent a little time on troubleshooting, but everything always worked out. So I don't expect a lot of trouble with this NIC swap.

Even so, I'm thinking about putting in the new NIC with the NVidia controller still enabled, and connecting my Ethernet cables to both. Then -- establish a new "Connection" with the new card before I decommission the old one. I actually think this will be a bit easier, even if only to replicate the proper IP, gateway and DNS settings from one to the other (obviously, I'd probably pick an IP for the new card that would be "old_IP + 1"). I wont' get started for a couple hours, so I'll check back to see if anyone has a brief comment, prohibition, encouragement, or thought.

The more I think of it, the two-NIC connection might cause more trouble. the new one should auto-configure on startup as soon as new hardware is recognized. Gotta think about it a little more . . . couple hours . .

 

[azazel1024]

IT should auto-configure correctly.

Should.

I've only seriously been in to networking for 15 years now. IMHO, networking in some manner or another (especially wireless) is probably the "hardest" computer related thing I've ever done and do with computers. Sometimes its flawless and easy, other times something simple is just days of headaches.

I totally understand on the initial capitol investment. I just know a lot of people who get recommended to use old equipment, especially NETWORKING equipment without thinking about the knock on costs.

Too much advice or people going "oh, this old HP managed switch is cool. Look at all the features I'll never need or use and it was only $50 off eBay!" without considering that that old first generation 2004 gigabit 48 port switch, of which they might use 10 ports on it ever, is burning 100+W and costing them $100+ year...where a 1-2 year old used gigabit semi-managed or dumb switch might use...10w and cost them $10 a year...and might cost the same or just $10-30 more than that old HP/Cisco/whatever switch.

I see it too sometimes with people using old P4s and stuff as PFsense boxes, when, especially for something like that, a super cheap mITX build with an extra 1 port NIC added to it might cost $150, if that much and might save $40-80 a year in electricity (or more).

Of course, sometimes you just don't have the money handy.

I've been slogging along with my "old" 2x1TB RAID0 array in my desktop, even though I am starting to push over 75% utilized on it and noticing the slow down, because I just can't justify spending ~$200 on a couple of new 3TB drives. Sooner rather than later it'll be at the point there I HAVE to upgrade...but until then...slog along.

 

[BonzaiDuck]

IT should auto-configure correctly.


Of course, sometimes you just don't have the money handy.

Or just other priorities. Old CPU and mobo notwithstanding, I'd think a big component of your power-consumption is the number of HDDs on a server. I became reticent about dual-3Ware 8-drive RAID5 or similar before I managed to think much about it.

It's enough to make you think about when to replace 1TB drives with 2TB or even 3TB, even on a one-at-a-time basis which you can do with drive pooling. But then, fewer drives means less file-duplication as insurance.

Still need to bring down the server and install that card -- today -- tonight -- whatever.

 

[BonzaiDuck]

FLASH!
FINAL REPORT: The Big E-net Cable-swap-out -- "Didn' hafta do Nuthin' . . . "

Shut down, unplugged, installed Intel Pro Desktop G-bit Ethernet Controller.

Booted up. Windows looks for the driver in some "cab" file, and comes up with the right one: "Intel Pro. . . Successfully Installed."

At this point, the Intel Pro is not cabled to my desktop switch.

Shut down again. Shut down my "bidnis" system and unplugged its RJ-45.

Plugged spare twisted pair to Intel Pro.

Booted up. Under "Network and Sharing Center" -- there are now two E-net controllers installed and working. In "Properties" for the Intel Pro's "Local Area Connection 2", all the values and sequences of the numbers I would have keyed in to the NVidia Controller are auto-reported in the Intel Pro's "Details."

"Uninstall" the original "Local Area Connection" configured for the NVidia Controller. Can't remember if I needed to uninstall the NVidia in "Device Manager," but remember answering the question "Do you want to uninstall the drivers as well?" -- which you choose, because you have them all in a "zip" file tucked away in a folder somewhere.

"start"-> "Restart" ->[disable LAN #1 Controller in BIOS, Save Exit and reboot.

Event log summary:

"System Log, beginning with the known marker of three "Event Log" info messages, has no red, and only benign warnings ("System is installing dll libraries . . yada yada. . ").

In the Application Log, the "LoService" for the Lights-Out Axonet Add-in reports "too many errors -- stopping service". Then the next info message: "LoService restarting . . . " followed by "LoService started successfully."

I'd bet money "Lights Out" will sort itself out with the next reboot. If not -- it can't be hard to fix: It's the only Add-in I haven't upgraded. And I know the license works for the new version.

 

[azazel1024]

Yeah, I keep my drive count low. 2x2TB RAID0 in my server and 2x1TB RAID0 in my desktop. Data is mirrored between them. I've got oodles of space on the server array, but the desktop is getting crunched. I also have a 500GB drive in the desktop that is around 78% utilized too. Sigh.

I try to keep drive count to no more than 2 for power reasons. I need at least 2 for performance reasons though. SMB3 multichannel and 2 links, so I can push a theoretical 235MB/sec...and did when the arrays were less full (desktop primarily at fault)...but I can still regularly get in the 210-220MB/sec range...just getting lower every week though. I figure a pair of 3TB drives in RAID0 should tide me over for at LEAST 3-4 years at my current fill rate, maybe longer, maybe not (a lot depends on 4k).

Fingers crossed I can make it to Black Friday and there are some nice deals. Newegg had some Seagate 3TB drives for ~$89 each this past year and I think they were even $79 each 2 years ago. I am hoping 5/6TB drives really hit the market in a big way between now and then so that lower capacity drives see their prices depress. I need to replace the desktop drives, but I am hoping to replace the server drives with the same drives at some point too. Then move the 2TB drives to a USB3 enclosure and I can add on 2TB drives as needed for capacity. Use it for secondary backup and hang it off something uber low power. Maybe that Bay Trail mITX board I mentioned (if any have regular DIMM slots, I've got a spare DDR3-1333 2GB DIMM laying around) in something cheap. just spin it up once or twice a week for backups.

Glad it seems to be working and hopefully the issue will resolve for you on the next power cycle.