HP DL140 G3 server death

[pnlrogue1]

We've got one of these and have an annoying problem -

 

We're running Smoothwall (Linux-based proxy/webfilter/firewall server) and keep getting a problem where one or more people will simply be denied internet access. After a lengthly process of fault finding, we found that the RAM had died (used the Memtest86 cd). Swapped the RAM, no improvement - it was the Motherboard. Swapped the board, fixed the problem. Board died again. Then again. Then again

 

All in all, we've had 4 new motherboards + the original and they almost all have had the same problem

 

Worst part of all is that the engineer spent ages installing the newest board, ran memtest and got the same error!

 

Has anyone else seen this before? Any ideas what we could do? It seems to be a fault which develops (normally no problem for months then it occurs). The latest motherboards definitely have newer BIOS's so sadly that's not it

 

Thanks

[DAckroyd]

sounds like something is killing the boards - have you fired up the board before installing it into the case - if so and it work and then when you put it into your case I would guess that something in the case may be shorting something on the board?

[pnlrogue1]

have you fired up the board before installing it into the case

 

No - too much of the hardware is part of the case to achieve that. Besides, it normally runs for months before the problem arises - this is the first time it happened straight away

[DAckroyd]

that sounds like a very odd problem... are all other componants working fine still and it's just the motherboard/ram that bring the problems?

[pnlrogue1]

Yeah, just the board

[DMcCoy]

We've got one of these and have an annoying problem -

 

We're running Smoothwall (Linux-based proxy/webfilter/firewall server) and keep getting a problem where one or more people will simply be denied internet access. After a lengthly process of fault finding, we found that the RAM had died (used the Memtest86 cd). Swapped the RAM, no improvement - it was the Motherboard. Swapped the board, fixed the problem. Board died again. Then again. Then again

 

All in all, we've had 4 new motherboards + the original and they almost all have had the same problem

 

Worst part of all is that the engineer spent ages installing the newest board, ran memtest and got the same error!

 

Has anyone else seen this before? Any ideas what we could do? It seems to be a fault which develops (normally no problem for months then it occurs). The latest motherboards definitely have newer BIOS's so sadly that's not it

 

Thanks

 

What is the error from Memtest? I've never found it to be very reliable at identifying faulty memory.

 

On the Smoothwall issue, then that is probably related to the guardian part, we have the same problem from time to time. Simply restarting the authentication and then the proxy usually fixes it. sometimes a reboot is required. How accurate is the clock compared to AD. The other possibility is that the users password has expired while they are logged in to windows, so Smoothwall will say that it is not valid (which is isn't) while windows will continue working until you logout.

 

I do not see how a hardware fault could possibly stop certainly users from logging in.

[pnlrogue1]

What is the error from Memtest?

It just crashed out at a certain point in the test. We know it's not the CD because after changing the board, it ran fully without error. Every time we have this problem, it crashes at the same point

 

Simply restarting the authentication and then the proxy usually fixes it. sometimes a reboot is required.

Tried that. All it does is shift the problem on to someone else - when we get the problem, we can't get rid of it. When the problem first arose, we spend weeks testing it with both our consultants and Smoothwall themselves - an update released last year to Smoothwall (the one where it massively rewrote the logging engine) was a result of Smoothwall looking at our system and finding that the log system was using too much RAM)

 

How accurate is the clock compared to AD.

Very close

 

The other possibility is that the users password has expired while they are logged in to windows, so Smoothwall will say that it is not valid (which is isn't) while windows will continue working until you logout.

Thought about that but it's not a user thing - same person can log on to another laptop and get full internet without needing a new password. In addition, changing the IP address removes the problem

 

I do not see how a hardware fault could possibly stop certainly users from logging in.

RAM errors like this can cause all kinds of problems - for the last 3 weeks, our internet connection speed was slow. Swapping the board may not have fixed the CACHE Access Denied error described in my original post, but it did bring our connection back up to speed

[m25man]

I would be looking at the PSU's and any add in cards fitted.

 

Had an Intel Server (A DC) fail the other week.

 

Changed the MoBo same fault!

 

Remove the Lights out add in card, Server came back to life (Luckily as the old board was still a toaster without the add in pcb)

 

So it was a peripheral board that killed mine.

Is it HP Ram or Clone? I doubt that HP would have entertained a board swap if it wasn't tested with HP Ram in it!

[pnlrogue1]

I would be looking at the PSU's

Never replaced the PSU's but if it was them, we would see the problem immediately after replacing the MoBo or within 24 hours, not several months later then all the time!

 

and any add in cards fitted.

Haven't fitted any - all the parts in that server are fresh from the factory

 

Is it HP Ram or Clone? I doubt that HP would have entertained a board swap if it wasn't tested with HP Ram in it!

HP RAM - they sent us new RAM to see if that's the problem but it made no difference. Tried changing the slots (along with every other possible combination of RAM - even one of the engineers tried!), same problem.

 

Everyone who's been involved agrees that it's the Mobo causing problems with the RAM - even HP

[Jamo]

Never replaced the PSU's but if it was them, we would see the problem immediately after replacing the MoBo or within 24 hours, not several months later then all the time!

 

 

Haven't fitted any - all the parts in that server are fresh from the factory

 

 

HP RAM - they sent us new RAM to see if that's the problem but it made no difference. Tried changing the slots (along with every other possible combination of RAM - even one of the engineers tried!), same problem.

 

Everyone who's been involved agrees that it's the Mobo causing problems with the RAM - even HP

 

I agree with the PSU, a faulty PSU can cause increased wear on a component by varying the voltage applied too much causing inconsistencies and processing errors.

 

If the case was the mobo with RAM then everyone with that model would currently be getting recalls.

[pnlrogue1]

If the case was the mobo with RAM then everyone with that model would currently be getting recalls.

 

That's what I was asking in my original post!

 

--- Edit ---

 

Will try asking HP if that could be a problem and see what they say

[DAckroyd]

I don't suppose you have a second one of these machines around do you? if so it might be worth swaping the hdd's between them and see if the problems move

[pnlrogue1]

I don't suppose you have a second one of these machines around do you? if so it might be worth swaping the hdd's between them and see if the problems move

 

Sadly not (I say 'sadly' but in reality, if I never have to work with one of these XYZing servers, it'll be too soon...)

[DAckroyd]

how old is the server? has it reached a point where you can say it needs an upgrade or where you can upgrade another server and move the old server into the DL140's place?

[pnlrogue1]

how old is the server? has it reached a point where you can say it needs an upgrade or where you can upgrade another server and move the old server into the DL140's place?

 

I was brand new when we bought it and started having problems within 6 months. At the moment, it is roughly 15 months old!

 

We are already setting up an old Acer Core2Duo desktop machine as a replacement

 

All I wanted to know was if anyone who had one of these servers was having similar issues or if we were just really unlucky!