Please help with dying server

[bossman]

@tmcd35:

 

Just use winrar or 7-zip to extract even though it is in a self extracting exe wrapper I personally like to extract to where I want it. It runs well across a network share also.

 

As you know it is just for testing purposes only if you like it then please purchase it.

;)

[Michael]

God knows. All three servers are the same. In all three cases the onboard NICs are bonded 2Gbps and are the only NICs actually being used.

 

If you don't need them, don't put them back in! Touchwood all will be well now. I'd rather test them in a spare workstation than a server.

[tmcd35]

Weeelllll, here's the thing (I love to overly complicate matters ) - I really want to move the school down the virtualisation route. As I said above these servers are a nigh on ideal starting point.

 

These machine that's playing up was to be (is to be) the first virtualised. The plan for October half term is to make a VM copy of the current running server then rebuild the physical box with Win2k8r2 Hyper-V.

 

I'd be looking at using atleast 4 NICs in each server (see above). So ideally I'd like to keep these in the server if they work.

[SYNACK]

If they are propper hp NICs they are likely to need a firmware upgrade too, put them in a workstation and upgrade them in that. This does not change the hardware identities and so there will be no driver concerns. Many other brands also support firmware upgrades on the higher end models that you would expet to find on a server.

 

One of the cards may be toast and so a firmware upgrade could help flush it out.

[tmcd35]

Unless I get a very unpleasant surprise in the morning this issue is now solved!

 

I want to thank everyone who contributed to this thread, you helped me sort out in my mind what I was doing, what I'd done and in some cases helped me view the problem from a different angle.

 

Okay, enough of the suspense - it was one of the NICs. No I have no idea which and I reliased (thanks Michael/Sysnack) that I pretty much don't have to worry which.

 

For what I want to do NIC wise in October I have at least two spare working 1Gbps NICs in the other two servers I can pinch and I have plenty of 100mbps NICs. Maybe one day I'll have a reason to test and find out which of the three NICs killed my server!

 

I downed the server at 1:30pm and put all three NICs back in. The system instantly blue screened. Took the NIC's out, no blue screen - windows boots. NICs back in and blue screening again.

 

Actually got additional information from the blue screen this time - NMI memory parity check error. But this only happens with the NICs in.

[mac_shinobi]

Actually got additional information from the blue screen this time - NMI memory parity check error. But this only happens with the NICs in.

 

would be interesting to know if a firmware update will clear that error message or if its an actual memory error on the NIC in question.

[tmcd35]

The NIC's are el cheapo Realtek 8169 1Gbps desktop NIC's. Tbh i can't be asked to find out if a Firmware update is even available for them. As it stands we have plenty of spare 1Gbps (and if necessary 100mbps) NICs that one (or more) of these three being faulty doesn't effect our future plans in anyway.

 

The only thought I've had is - could the problem be with the PCI riser rather than one of the NIC's themselves. That would cause problems in October...

[mac_shinobi]

The NIC's are el cheapo Realtek 8169 1Gbps desktop NIC's. Tbh i can't be asked to find out if a Firmware update is even available for them. As it stands we have plenty of spare 1Gbps (and if necessary 100mbps) NICs that one (or more) of these three being faulty doesn't effect our future plans in anyway.

 

The only thought I've had is - could the problem be with the PCI riser rather than one of the NIC's themselves. That would cause problems in October...

 

Not sure what a PCI Riser is unless you mean the PCI Slot ?

 

If im honest I would go with Intel for the network cards and ditch the realteks

[SYNACK]

Could be the riser but it could just as easily be one of the cards. The only way to test would be to put them in a different server and try. If I recall correctly there is not that much hardware in a 380 riser.

 

I would also look at ditching the realtec cards if you are going to put heavy usage over them as they will hit the CPU quite hard in comparision to fully accelerated cards. For stuff like iSCSI or heavy traffic/trunking this will burn lots of CPU cycles that would be far better applied to serving client requests. Something with full TCP/IP offload and - for iSCSI - iSCSI acceleration/HBA will end up providing you with a much better end result.

 

Remember that you can only team like network cards with the vendor apps so you would be able to trunk the two integrated ones with the hp app but would need to instal;l teh intel app to trunk the intel ones. This can also cause a bit of strife when the intel and hp teaming solutions get into an argument (read bluescreen hell).