Long Win 7 boots with RAID drives connected

[Sabrewings]

So, I was barking up what I think was the wrong tree, but you can see my previous post on this here:

http://forums.anandtech.com/showthread.php?t=2437554

The gist of it now is that with my four RAID drives connected, my time spent on the Win 7 splash screen (where the four dots come together to make Windows logo) is around 3-4 minutes.

During this time I can tell by my MB's LCD output (which displays MB socket temp once POST is complete), my CPU is around 70-80% load. The HDD light (which only functions for the RAID array since my OS drive is a PCIe SSD) pulses occasionally in a rhythm during this entire time.

If I remove power to the RAID stack, it boots fine. Previously, I backed up my data, broke the RAID, updated BIOS, rebuilt, and reloaded the data. Things were fine until I went and made a system image backup with the built in Windows 7 backup utility. After that it went back to the long boots.

When things are working right, I usually don't even see the four dots flash together to make the flag it goes through so fast. To have to put up with a nearly 5 minute total boot time from button press to logging in is pretty annoying.

Once the issue came back, I said screw it for a bit since Win 10 is around the corner and I'll probably fresh install to it once I have an image in my hand, hoping that fixes the issue.

If anyone has any insight I'd appreciate it. I can't really tell at this point if it's a Windows problem or a hardware issue. Other than this, everything is fine and damn snappy on my machine.

BTW, before I get any lectures on using Greens in a RAID, I know what I'm doing with them. They're essentially Reds with TLDR disabled. I have link power management disabled and I disabled the Idle3 timer, and if a bad sector comes up it'll just time out that drive and kick it from the RAID at which point I'll replace it. They don't cause any problems and I doubt being Greens is the problem in this thread because plenty of users on FreeNAS and NAS4Free use configurations like this, admittedly with a different OS which shouldn't matter.

 

[Ketchup]

So is Windows backup pointing to the RAID array in some way? Are event logs showing the same error you mentioned in your other thread? Is the RAID array in the Intel chipset? Is there a chipset update (Windows or manual) that occurred around the time of this issue?

 

[AnonymouseUser]

BTW, before I get any lectures on using Greens in a RAID, I know what I'm doing with them. They're essentially Reds with TLDR disabled. I have link power management disabled and I disabled the Idle3 timer, and if a bad sector comes up it'll just time out that drive and kick it from the RAID at which point I'll replace it.

I won't lecture you, but I have a Green drive, and I hate it (I also disabled Idle3). I plug it in, make a backup, unplug it, and it sits on my desk in it's bubble case because it's extremely sluggish.

That said, in your case I would test just three drives at a time until you've tested with each drive removed from the array. That should tell you if one of those drives is the culprit.

 

[Sabrewings]

So is Windows backup pointing to the RAID array in some way? Are event logs showing the same error you mentioned in your other thread? Is the RAID array in the Intel chipset? Is there a chipset update (Windows or manual) that occurred around the time of this issue?

Yes. That's where I stored the image when I made the backup and started noticing it for the second time after that. I was making backups to it when I noticed it before. To test the theory, I deleted the backup and completely disabled all Windows restore and backup options. The problem persists.

I won't lecture you, but I have a Green drive, and I hate it (I also disabled Idle3). I plug it in, make a backup, unplug it, and it sits on my desk in it's bubble case because it's extremely sluggish.

That said, in your case I would test just three drives at a time until you've tested with each drive removed from the array. That should tell you if one of those drives is the culprit.

Single drives I can imagine would be terrible. I have four of them in the array and I get about 480MB/s read and write speeds, so it's not too bad for general archival use.

As for testing them, that's what I was starting to do last time when I broke the array. I suspected one was bad. After rebuilding the array it just stopped doing it. I suppose I could pull all of the data off again and not break the array to keep the rebuild times as fast as possible. As it is a rebuild takes nearly 48 hours.

I really don't think it's a drive. I was fine for a week after breaking the array then it started happening again, which points to a software problem. I might try making the drive inactive in Windows and see what happens. Or, I could break it and just make backups to an external as needed.

I'm hoping a fresh format for Windows 10 fixes it. I never had this issue at all on my last PC running a RAID 5 under Windows 7.

 

[Ketchup]

You could test each drive in the array individually with a bootable CD. Seatools, for example. The fact that the drives are fine with a clean write and lose speed over time could mean there is a bad spot on one of the drives.

 

[Sabrewings]

You could test each drive in the array individually with a bootable CD. Seatools, for example. The fact that the drives are fine with a clean write and lose speed over time could mean there is a bad spot on one of the drives.

They're not losing speed. The boot is simply hanging with high CPU use and the hard drive access light occasionally pulsing. The OS is not loaded on the array. I boot from a PCIe SSD. The HDD light can only correlate to the RAID because I have no other HDDs and my PCIe SSD doesn't trigger it.

I do like the Seatools idea though. Could eliminate any potential issues in hardware.

 

[Ketchup]

When the problem went away, did you wipe your OS, or just the array? Just wanting to make sure I understand the events that transpired.

Do you have any software managing the array (Intel Matrix Storage Manager, etc)?

 

[Sabrewings]

When the problem went away, did you wipe your OS, or just the array? Just wanting to make sure I understand the events that transpired.

I just broke the array. No other software changes. Rebuilt the array the loaded data back in.

Do you have any software managing the array (Intel Matrix Storage Manager, etc)?

I have the Intel RST software. It reports everything as green with the drives and array, even after SMART tests.

 

[dionasaur]

You could test each drive in the array individually with a bootable CD. Seatools, for example. The fact that the drives are fine with a clean write and lose speed over time could mean there is a bad spot on one of the drives.

I suspect a bad sector as well.

 

[Ketchup]

I suspect a bad sector as well.

Either that or the Intel RST is having trouble with the array once loaded with data.

 

[Sabrewings]

I'll test them individually and see what I can come up with.

If it is a bad sector, why does it even affect my boot process when no boot data is stored on the array? And why would I be seeing max CPU usage at this time?

 

[dionasaur]

I'll test them individually and see what I can come up with.

If it is a bad sector, why does it even affect my boot process when no boot data is stored on the array? And why would I be seeing max CPU usage at this time?

Despite not being bootable data, I think the only clear explanation I can give is that any "bad" or "corrupt" data greatly hinders performance. It also could be Windows trying to index the information on those drives which is causing long delays.

I.e. Outlook performance issues with long holds / freezes are usually caused by corrupt OST files or corrupt email data.

Also, I have a question. When you have long holds, what screen are you at? (I apologize if you stated this somewhere in one of your posts)

 

[Sabrewings]

It hangs at the Windows splash screen (black screen with logo) for about four minutes.

Also, I have turned off any indexing functions for the RAID volume.

 

[Sabrewings]

I deleted the RAID and it is currently booting very quickly like it should. The issue is definitely RAID related.

I'm benchmarking and investigating each individual drive for right now, then I'll build the array back and reload it with data. I'll try to get an idea for exactly when it slows down again, if at all. I'm not finding anything wrong on the hardware side (all drives perform well and equally when not in RAID), so I'm not sure what else to do here.

 

[Ketchup]

Either that or the Intel RST is having trouble with the array once loaded with data.

I want to edit this, because it could also be something to do with Windows Backup, if the drives come out of testing as perfectly healthy drives. It would be a good test to throw some data on the array after recreation, leave RST and Backup out of the picture, and see if thing hold up for while.

 

[Sabrewings]

Ok, I removed RST and we'll see where it goes. No backups for a while. The array is going to be initializing for the next day or so, and I'm going to let that finish before I move my data back in again.

 

[Sabrewings]

Little update. So far so good. I let the array fully initialize then loaded the data onto it. Still boots very quick. I'm not going to touch it otherwise for a week and see if it holds.

*crosses fingers*

 

[Ketchup]

Good. And is Intel RST running?

 

[dionasaur]

::thumbs up:: Hope all goes well

 

[Sabrewings]

Good. And is Intel RST running?

Nope, uninstalled it.