bad HDD diagnosis

[rallynavvie]

Wanted to bounce some ideas off the other IT folks here before dropping $200 on a new boot drive:

Symptoms: Came home to find my main workstation had rebooted and was frozen on the Windows loading screen. Hard reset and selected boot Windows normally. During the Windows loading screen the machine rebooted on its own. During this time I notice the HDD light comes on solid but I'm not feeling any drives accessing while I've got my hand on the HDD cage. I disconnect all but the boot drive and I'm still having a hard time getting into Windows. Once in it takes forever to load everything. Added other drives one by one and they don't affect the stability at this point. When opening multiple programs (for instance opening Firefox while watching a movie) it takes forever.

Troubleshooting: SMART from BIOS reads as OK, though I've yet to see SMART do anything on HDDs. Chkdsk shows some errors which I believe it corrected. Sustained transfer test showed the drive is running a little slow but not that different than before (just timed a large file swap between it and one of the other drives). Pagefile is on the boot drive but I haven't gone to the extreme of moving it off to another drive. Ran O&O to defrag the volume.

Anything else I missed? PSU seems to be fine as I spiked the GPU to draw as much power from it as I could and nothing flinched. Memory is an unlikely culprit as once things are loaded they seem stable and my memory is some of the best ever made. Board is a little flaky with a couple of bad caps but they've never acted up in this manner before. Anything else I can use to test the access ability or memory buffer of the drive? I'm thinking that may be the culprit.

Sad thing is this drive is still under warranty, a WD 74GB Raptor. Unfortunately I can't wait for downtime so I'll have to get a new 150GB model to replace it and wait for the other one to come back for some other machine to use. And my little Seagate 15k SCSIs just keep chugging along, 5 years and counting...

[Soopafresh]

http://benchmarkhq.ru

Tons of different tests.

I've used the 37GB WD 10K RPM models and I didn't notice a performance boost. Are you sure you need another one ?

[redwudz]

JMO, but you might have dismissed memory too quickly. You could run Memtest86 on it: http://www.memtest86.com/ Boot it off a floppy and run for a few hours at least. Though I've had better luck just trying a different RAM module. I have had strange intermittent problems from memory.

I've had S.M.A.R.T. detect failing drives several times, but it's also missed a few that died. But probably replacing the HD is a good idea. Then you either have two good drives or you solved the problem.

From what you've checked so far, and if the RAM is OK, I'd guess the next possibility is the MB. It might have a problem, but they rarely seem to be intermittent. Usually they just go quickly. I'm assuming also you checked the temps and voltages in BIOS.

And the most basic, reseat and replug all connectors and boards.

[oldandinthe way]

What does CPU utilization look like once you're in? Is some task eating the CPU causing the lag.

What are you using for antivirus? Could be an infection which is invoked at boot.

Have you tried safe mode?

Notice I have not suggested any hardware problem. You do not have an error which points in any one place. Could be hardware but the symptoms are such that the only way you'd localize is to replace parts by trial and error.

Given the testing you've done - I'd troubleshoot the software, and consider rebuilding the boot drive. If you succeed in rebuilding your problem was definitely not hardware.

[rallynavvie]

CPU is nil during most of these tasks, just the usual utilization. My task manager is always minimized to the system tray so I can monitor CPU usage. I've been doing this since my first Win2k build.

Safe mode results in similar lag when trying to open applications. I had to boot into safe mode to run some adware scans and and AV sweep.

Reseating connectors was the first thing I did after powering down

The performance boost of the 10krpm SATA drives isn't sustained transfer speeds, it's mostly in the response/seek times that I notice a difference. It isn't much but it is more noticeable on an SMP platform where apps are being opened and closed ad nauseum. Scratch space is generally on my SCSI stripes but I moved the pagefile off of that volume because it got annoying listening to those drives access every 15 seconds. Loathe to put the OS on that 36GB RAID 0 volume.

I think I'll load up the last image I did about a month ago and see if the slowdown is still there. If it is I'll be almost 100% certain the issue is hardware related.

[oldandinthe way]

I've never seen it on a hard disk controller but any chance DMA is diasbled. If it is it wouldn't necessarily show much difference in sustained transfer as long as it didn't run out of CPU but it sure as heck would slow your boot process and swapping.

[aedipuss]

run WD Data Lifeguard Diagnostic for Windows

http://support.wdc.com/download/?cxml=n&pid=1&swid=3

[fritzi93]

I've never seen it on a hard disk controller but any chance DMA is diasbled. If it is it wouldn't necessarily show much difference in sustained transfer as long as it didn't run out of CPU but it sure as heck would slow your boot process and swapping.

I was hesitant to ask that question for fear it sounded too ridiculous. But that happened last week on my computer. It was lagging badly, did malware scans, ran system file checker, page file is on a different drive, no suspicious services running. Ran every damn diagnostic I could think of on the hard drive and memory until, not knowing what else to do, I checked the IDE channel.

But I still don't know the underlying cause. Turned system restore off, then back on. The problem has not recurred for several days now, but I'm still at a loss... :P (Yes, I have backups and a couple cloned drives, but I'd like to understand this.)

Sorry for getting slightly away from the original question. I'll be very interested to learn how you resolved your problem.

[oldandinthe way]

Any hardware function controlled by software can be subject to the usual causes of problem, bugs, malware, system crashes.

If the OP's problem is the same as yours, he may discover his backup will restore functionality, and neither we nor he will know if DMA is off.

[rallynavvie]

Have you ever used the WD Data Lifeguard tools? It has a walkthrough on how to install a hard drive and you can use it to check serial numbers but otherwise it contains all the same things you can find in Windows itself. The app is pretty useless.

I booted into recovery console from my MCE disk and the HDD light stayed lit. Chkdsk wouldn't run saying there were unrecoverable errors on the disk. Swapped out SATA cables to the disk and there was no change. I'm re-imaging now.

FWIW DMA was enabled and all other settings in BIOS looked fine. I run my PCs behind UPS to prevent weird power anomalies from glitching the BIOS (it seems to happen most often during brown- and black-outs).

[aedipuss]

not the data lifeguard tools - the Data Lifeguard DIAGNOSTIC

different program.

[rallynavvie]

Re-image didn't help at all so I went and picked up the new drive. I don't think the old image would have worked with the larger drive so I just did the installation from scratch and will knock out a new image later this week once I've got everything updated to where I want it.

Now how do I re-assign drive letters to everything? When I installed new I forgot to add drives as I needed them and messed up all my drive letters. In particular my boot drive is now listed as H

[fritzi93]

Now how do I re-assign drive letters to everything?

Administrative Tools -> Computer Management -> Disc Management. Right click a drive in top right pane -> Change Drive Letter and Paths. For optical drives, scroll down bottom right, right click, etc., etc.

[oldandinthe way]

Not talking about whether the BIOS has DMA enabled, interested in whether Device Manager shows it as enabled. The hardware can be enabled but Windows doesn't use it if not enabled in device manager.

[fritzi93]

Not talking about whether the BIOS has DMA enabled, interested in whether Device Manager shows it as enabled. The hardware can be enabled but Windows doesn't use it if not enabled in device manager.

Yeah, saw nothing untoward in the BIOS. It was only when I looked at the IDE channel in Device Manager that I saw it (the OS drive) was in PIO mode.

[rallynavvie]

Now how do I re-assign drive letters to everything?

Administrative Tools -> Computer Management -> Disc Management. Right click a drive in top right pane -> Change Drive Letter and Paths. For optical drives, scroll down bottom right, right click, etc., etc.

I know that much, but you can't re-assign the boot volume from there. I thought you had to boot into recovery console or something.

[aedipuss]

you can change the drive letter with regedit but the system will become unbootable. at which point it can sometimes be fixed by booting the install cd into recovery mode and rebuilding the hive. much easier to simply reformat and reinstall. this time remember to unplug all other hard drives before installing winXP.