AHCI quandary. – 257 days old

[MSUredux]

I mentioned this in passing in my thread bitching about D2D. There is now more to the story, so I'm restating the issue here.

OK, I have 4 SATA HD's and a SATA Blueray/DVD drive in my main machine. The MB has 6 SATA ports controlled by the Intel ICH10 Southbridge chip. And 2 SATA ports controlled by separate Gigabyte SATA2 chip. Now the Gigabyte has always been disabled and everything was on the Intel controller. I had it set to run in AHCI mode.

Last week my main HD and another one stopped being recognized by the BIOS when in AHCI mode. The Windows install became trashed because of it. I tried every drive in every Intel port and the three that were working were recognized in all of them and the two that weren't recognized didn't work in any of them.

So I switch to IDE mode and all the drives were recognized. I then reinstalled Windows.

On a whim I thought I'd see if for some strange reason they would be recognized in AHCI mode again. So I enable the AHCI driver in Windows and proceeded to reboot. Switch to AHCI in the BIOS and low and behold, my main drive was recognized. But not the other one.

Now another on a whim moment. I enabled the Gigabyte SATA chip, set it for AHCI, and connected the still offending drive to one of those SATA ports. Guess what? It was recognized!

So now all the five drives are back to AHCI mode. But one of them is on a separate controller. To recap, at this point I have three HD's and an optical drive that are recognized on eight different SATA ports when in AHCI mode. And one HD that is only recognized in AHCI mode on the two SATA ports on the seperate controller.

So if anyone was able to follow this long dissertation, I'm wondering if anyone has a clue what happened.

edit: Oh, and the reason I thought to check to see if for some reason the offending drives would miraculously be recognized again in AHCI mode is that there was a noticable slowdown in boot times and program launch times which were unacceptable to me.

[ZemaTalon]

When it's on the Intel controller, is it not seen by the BIOS, or only not seen by the OS?

[MSUredux]

Steve, it's not seen in the BIOS when on the Intel controller. In fact, it hangs when trying to detect it. So I had to disconnect the drive to even get into the BIOS to change settings. Really odd because I've had it set up this way since Windows 7 came out. I could see if something went south on the MB and none where recognized. Or even one port go bad. But to have all the other drives be recognized on all the Intel ports and this one drive to all of a sudden not be recognized by any of them (but to be recognized by the Gigabyte controller) is really odd. And perhaps even more odd is the one drive that all of a sudden decided to be recognized again.

I'm not too horribly concerened at the moment since it's all working in AHCI mode now...but what if other drives start to exhibit the same behavior? There are only two Gigabyte ports.

[Grogan]

A flaky drive exhibits flaky behaviour and another SATA controller may or may not choke on it the same. I had a strange one with a Western Digital Scorpio Black 2 TB drive. It would hang for several seconds while being detected by the Intel ICH10 AHCI controller. (I never thought to try it it any other mode, for I only use AHCI. I don't even add support for anything else to my Linux kernels). Then, often times, it would block i/o for a split second. This was mostly noticeable in Linux, for I had my main OS installed to that drive, as well as a storage partition. It was also the drive I have a 1 TB Windows partition on, dedicated for Steam and I noticed that some (not all) games were freezing for a split second. Usually not enough to cause harm, except I was getting creamed one night in F.E.A.R. playing with Zema. I couldn't get a shot off in most encounters. I didn't even make the connection between the two at first (Linux behaviour and those games), blaming other things for the subtle problems. I tried to figure it out in several ways, at one point thinking my SATA controller was flaky and then thinking that both of those drives just weren't getting along together on the same controller. It didn't seem like the drive was bad, and it passed diagnostics (natch... piece of shit subtle problem that it was)

It wasn't until later that I realized it was a faulty hard drive. I moved my Steam directory over to another drive, and sure enough, the milliseconds of freezing in those games stopped. I quickly went to the store and bought another one, a 2 TB drive the same, but a newer model. I transferred my Linux setup and Steam directory back there and everything was marvelous. Sent the drive back and got a replacement that is doing fine in another computer storing backups. (1 big 2 TB, Linux Ext4 partition)

Anyhow, now that you have it connected, run WD diagnostics on it. It will at least tell you if the S.M.A.R.T. data is kosher.

[MSUredux]

Good idea to run the S.M.A.R.T. diagnostics Grogan. All the drives are Western Digital Black's. Except for the one now on the other controller. It's a VelociRaptor.

[MSUredux]

Well they all pass the WD diagnostic tests.

[ZemaTalon]

Have you tried it in another machine? It'd be interesting to see how it behaved on a 3rd controller

[MSUredux]

Hmm, hadn't thought of that. It's kind of a pain in the ass to get out of this machine. I may just have to do that though when I get the chance. My secondary machine is running in AHCI also, so it wouldn't be an issue testing it there once it was extracted from this case.

It's an original Antec P180. Tons of room...except for cableing four HD's.

[Grogan]

Western Digital diagnostics is actually pretty awful. It's designed to "pass drives" not to find faults

(I do like it for the reason that it's a very fast bad sector check, apart from the SMART data checks. It will find blatantly faulty drives very quickly, but is useless in marginal cases. Who has time for exhaustive surface checks that take several hours? If time was money, and it is to me, that would cost 2 new hard drives)

[MSUredux]

Ready for a new wrinkle? I was going to take out the offending drive and try it on my other computer. Since I was in there I thought "What the heck, since it's been working on the Gigabyte controller under AHCI, let's plug it back into the Intel controller and see what happens with AHCI there". Would you believe the SOB was recognized? I proceded to disable the Gigabyte controller and the drive's been working fine on the Intel controller for the past day. I hate these kind of problems.

[ZemaTalon]

I bet it's got you wondering what's coming next Yeah those intermittent problems are the worst. If it fails again that might be a good time to test it on another machine. And if it works on the other machine, I'd probably leave it there a while to see if it eventually acts screwy there.

[Grogan]

Cable connection? (It may have felt fine, but sometimes you have to physically remove and reattach connectors or flex screwy cables a certain way before they'll work, though I haven't see the latter occur with SATA cables themselves. Connectors are finicky though.)

[MSUredux]

I don't think it was the connections, because I had disconnected and tried every drive in every port. Plus the cables are the kind that have the locking clip on them. I don't know. I guess I'll just have to wait and see...