Server lacks up Hard Drives thrashing

[spliterman]

I have been experiencing an issue lately. The SBS 2003 server will "lock up" and the lights on the hard drives are flashing pretty fast. When this happens, I cannot log into the console and all clients lose connection to the network. I cannot find anything in the log files that occur around the time this happens. I have seen this before when a client tries to copy a large file from his computer to the server, but I do not think this is what is happening this time. Has anyone experienced such an issue?

Thanks

 

[noveyron]

What do the event logs tell you?

Does SBSBPA tell you anything important?

What hardware do you have? Make, CPU, RAM, HD space?

Any non standard applications?

How often does it happen, is there commonality between incidents?

How do you currently resolve the issue, hard reboot?

 

[wahnula]

If there are no event logs then I would look at the hardware. Maybe a worn/underpowered PSU, overheating CPU/RAM, etc. Next time it locks up, reboot into BIOS and find the temperature monitor feature, see how hot the CPU and mainboard are. If you don't have that feature download and run SpeedFan and monitor the temps.

It could be just about anything, I've even had a bad video card lock up a PC, but at least eliminate case & CPU temps as suspects.

Tony

Users helping Users...

 

[spliterman]

Thanks for the recommendations! No I do not find anything in the event logs around the time of the lockup. When this happens, the hard drive lights flash almost constant. To resolve the issue, I disconnect both network cables so there is not any traffic to the server then about 10 minutes later, the server calms down and everything is fine again. Plug the cables back in and all clients connect like nothing ever happened. There is absolutely no commonality between incidents. I ran the SBSBPA and it comes up clean. The server is a custom build about 5 years old (this may be the problem) with dual 3 Ghz processors and 4 GB RAM. The C drive(system files) has 4.75 GB free and the other drive (files) has 200GB free. Any good tool to test the condition of the hard drives? Since I do not think the server has ever been defragged, could this be the issue? Thanks

 

[noveyron]

Hardware looks good, I would say that you should initially avoid a defrag, given that you are running Exchange (depending on how you have it configured, obviously).

Your hard drive manufacturer should have tools available on their website for testing the integrity of their HDDs. I wouldn't imagine that this is an issue, although a chkdsk /r might be worth a run.

You might also want to have a look into deploying some monitoring tools:
Poolmon

http://support.microsoft.com/kb/177415

Procmon

http://technet.microsoft.com/en-us/sysinternals/bb896645.aspx

Perfmon

http://www.microsoft.com/downloads/...98-c3a1-4644-9622-faa046d69214&DisplayLang=en

 

[spliterman]

noveyron, thanks for your advice and the links. Yeah, I was wondering about the defrag myself. Has anyone ever experienced such an issue on SBS 2003 when copying large files? If I copy say a 1 GB file from my client to the SBS or even from a local folder to another on the SBS, the machine seems to "freeze" all processes (exchange, clients disconnect, etc) until the copy is complete. This cannot be normal for SBS, but I cannot seem to pinpoint where the problem lies.

 

[ShackDaddy]

Do you have a single RAID5 array? How are your disks configured?

Dave Shackelford
ThirdTier.net

 

[spliterman]

Yep-- RAID 5 with C drive for system (24.4 GB with 4.15GB free) and drive E (441GB with 233 GB free)

 

[technome]

Been working with raid since 1989, 99.5% of the time the raid is not the issue, hard disk errors repairable by chkdsk, can be an issue, as could drive firmware issues. Assume you ran chkdsk, drive firmware issues which cause drive thrashing then a lockout are extremely rare. Assuming you have a hardware raid, most fairly recent raid adapters place no more then 5-9% load on the system under a heavy load. Most of the time with hard disk thrashing,it turns out to be caused by network delays or a mis-behaving program,lack of drive space, memory leaks/resource exhaustion.

Do you have a managed switch, so you can monitor port errors ?

Agree with noveyron, run procmom and pool monitor. Open Taskmanager sort by memory usage, check if an process continue to increase in memory use, same for handle count.


........................................
Chernobyl disaster..a must see pictorial

http://www.kiddofspeed.com/default.htm

 

[spliterman]

Thanks to everyone! I think I may have found the culprit. I have been monitoring this server closely all day. The hard drive lights started flashing very fast almost staying solid. After a quick look at the processes, I restarted the IIS Admin service and the lights calmed down. Now I just need to figure out why this is happening. Thanks again.