Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

Hello to every body and greetings from Greece

I have a rather unusual situation and i am running out of ideas. I have this old server (IBM x205 - P4 2.4Ghz,3xSCSI 36gb) and about a year ago I decided to use it as an additional domain controller and fax & file server. For this task I had a Delock 70154 SATA card along with 2x320 SATA II HDDs

Everything was going super smoothly until about 3 weeks ago. I was on a trip and I was informed when I got back that the server was found frozen. Well, I considered it was a glitch since a simple power down / power up fixed everything. Again, two weeks ago another freeze situation. It got suspicious, but again after a power down / power up everything was running.

Again it froze and when I powered it up it came with a message that the domain services could not start due to NTDS corruption. Booting in safe mode revealed that there was an issue with the SATA Raid (degraded).

After a lot of searching I degraded the server, cleaned Active Directory, pulled both HDDs out (one of these was really BAD) and recovered my files (I had some problems with ho Delock handles the redudant HDD).

Right now my server is vanilla simple with only what the factory installed. Here is where the fun begins. Everyday when I arrive at the office I find this particular machine dead and I mean totally dead. Just a black screen and nothing else. The CPU fan is working, the PSU is working. Keyboard and mouse are dead (they also lock my KVM) as is the network. The machine is DEAD.

I power it down forcefully and then power it up. For the 8 hours I am in the office it works, either idling or running some kind of diagnostic. When I leave the office, after some time (maybe after 30 minutes, maybe after 4 hours) the machine dies. This is the information the event log shows

"the previous shutdown at xx:xx:xx was unexpected"

I have tried the following: Memtest: nothing showed up Passmark burn in test: nothing showed up Carefull study of the event log: nothing of interest that I can see Set the option to not autostart after a crash so that I could see BSODs: nothing to see Power scheme is set to never sleep.

I know there are a lot of other tools that heavily stress a machine like OCCT but the machine is old. Today I will give them a try nevertheless. One idea is to reformat it, but I really like to find what is causing this because I could get to a situation that everything is working for a while and then kaboom, one day again it is dying. I really need a helping hand and every opinion / idea is welcomed. I know the obvious solution is to never leave the office but I have a life. Sorry server. :)

P.S This situation with the machine dying has been going on for about one week. Everyday I would set either the RAID to rebuild or to copy/recover files and while everythig was working

share|improve this question
2  
sounds like one or more pieces of hardware are going... I wouldn't waste much more time messing with it and replace it, but that's just me. –  SpacemanSpiff Dec 30 '10 at 6:24
    
Greetings @tsgiannis! I cleaned up your post to make it easier to read. I attempted to keep your original intentions in tact. Hopefully my edits will make it more likely that people will read and then answer your post. –  Wesley Dec 30 '10 at 7:15
    
This is freaky. I have encountered situations similar to that of yours and I would sit there like Sherlock Holmes with a pipe in my mouth (or mint in yours, if you prefer) pondering about whom have I offended bad enough for them to put a hex on me. I believe the kind folks here have an answer for you that will show you the way to enlightenment. :-) –  Michael Feng Jan 24 '11 at 6:38
    
How long did you run Memtest? Anything under 24h doesn't count ;) –  mad_vs Jan 24 '11 at 9:24

5 Answers 5

Sounds like a very old server. Also sounds like it is dying a slow and painful death. There is a reason why most IT departments replace their hardware after a certain period of time. The reason is because it costs far too much money to support old hardware. I'm betting that the server you have is 6-8 years old. Get rid of it. It will be more trouble than it's worth. You probably have a bad motherboard or a bad processor. You probably won't be able to find either one new so you'll have to buy used parts. With the used parts, you don't know if you're introducing new problems into your server. Like I said, just get rid of it. It's not worth the effort/time/money you're going to have to put into it.

share|improve this answer
    
Well i know is a really old server ...but the task of fax /file server is way too low to consider a new server just for this. –  tsgiannis Dec 30 '10 at 6:57
    
And one more,when i say file server i mean just some shared folders where users dump their files....nothing critical....(almost) –  tsgiannis Dec 30 '10 at 6:58
1  
What's your time worth? More than that, what's the time of the employees at your company worth? If they can't access files on a file server, I'm guessing they can't be as productive. A brand new Dell T310 with a 3 year hardware warranty starts at $550. Even with a Celeron processor, I guarantee you it's better than the old IBM you have. I don't think that cost is all that hard to justify. –  Jason Berg Dec 30 '10 at 7:24
    
Well right now i am running on dead time...due to inventory. –  tsgiannis Dec 30 '10 at 7:55
2  
@tmow There something to be said for the sin of wasting one's life and their employer's time by keeping old hardware that's past it's prime alive. It's a danger to both one's career and the livelihood of their employer. More stressed out people means more substances used to manage that stress and more problematic relationships. A server in a garbage dump has less environmental impact that a single stressed out human or an entire struggling company. –  Wesley Dec 30 '10 at 22:55

Well i found the ghost.....it took some time but eventually i got it. The problem was due to leaked capacitors...one of them decided that it is time to stop wasting my time and popped its fluids out... So hopefully its time for something new....:)

share|improve this answer

If it's more than around 3 years old, dump it - if not, use the warranty and have it repaired.

Troubleshooting hardware is for hardware modding enthusiasts - there's no valid business case for doing it on this level. Even the simplest of IT services should be powered by reliable hardware or the service won't be reliable.

The decreased power draw of a more modern x86 server alone would make a purchase of a new low-end server valid if you do the math (unless your power is somehow free).

share|improve this answer

I'll try to do some statistics for cpu and mem usage, remotely. I am the kind of person that gives another chance to the old hardware.

So, do some statistics, maybe your processor is overheating. Clean the heatsink, replace the thermal paste and see.

HTH

share|improve this answer
    
Well i have just installed...it has two suspicious temps around 125 C but i guess is a wrong sensor read.....everything in the case is cold....the CPU is around 38C...just something new....how to measure the temps of my SERVE RAID HDDS...they feel hot.. –  tsgiannis Dec 30 '10 at 8:38
    
Correction...i have installed speedfan 4.42 –  tsgiannis Dec 30 '10 at 8:41
    
Some people didn't liked my post, but, the topic is on how can the server be still used and not about how it can be replaced. My second suggestion is to test the hardware using IBM's diagnostic tools. –  Paul Dec 30 '10 at 9:21
    
well ibm doesn't have at least to my knwledge some kind of diagnostic tools.Dynamic System Analysis gathers info but i can't find any diagnostics –  tsgiannis Dec 30 '10 at 10:32
    
www-947.ibm.com/support/entry/portal/…, assuming is a "IBM xSeries 205" –  Paul Dec 30 '10 at 10:45

As I stated in one comment I'm agree that is interesting to fix the issue, to avoid unneeded garbage and to learn new lessons for the future.

An old server like this can still serve really well for less important stasks, like you are doing as AD domain controller, maybe it's a bit extreme using it as file server.

It came really handy to me using an OpenSUSE installation CD just to use the disk and ram tests. Test the hardware for errors is really important.

Then I'd really consider cleaning the contact of the RAM (I normally use a rubber that is cheaper then some products).

share|improve this answer
    
Active Directory a less important task...? It's only critical to just about everything! –  Ben Pilbrow Dec 30 '10 at 11:57
    
I've used the wrong word, I meant that is something that you can substitute easily with an other one in case of hardware failure (much more a secondary domain controller), it's not that easy with a file server. Of course is a critical service! –  tmow Dec 30 '10 at 12:28
    
Well last thing before format...change PSU(??) –  tsgiannis Jan 4 '11 at 11:31

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.