Various drives lose their partitions repeatedly!

[Amby19]

This is truly weird! But before I go into details, please note that I do not need help recovering the data from the missing partitions. What I need is help in trying to figure out what's going wrong and fix it.

I'm running XP Pro/SP3 on a home-built box with an ASUS P5WDG2 WS Professional mobo that has worked great for years. But starting recently, all the partitions on some disk drives have repeatedly gone randomly missing!

Here's why I say "repeatedly": I use Acronis Disk Director 11 to recover the lost partitions, which works fine (but takes hours since it has to parse through 1 or 2 TB to find all the partitions' signatures). But a few boot-ups later, all of the partitions are gone again. It's like a loop, except that it's not always the same drive -OR- controller!

I started out with 8 hard drives on this high-end workstation (all partitions are NTFS):

[ALPHA]

: 3 normal, non-RAID SATA-II drives connected to the standard Intel ICH7R disk controller on the mobo.

[bETA]

: 2 SATA-II drives connected to the on-board Marvell 88SE6145 RAID controller in a RAID 0 configuration.

[DELTA]

: 1 normal (i.e., non-RAID) SATA-II drive also connected to to the on-board Marvell 88SE6145 controller.

[GAMMA]

: 2 SCSI-320 15K RPM drives in a RAID 0 configuration (which hold my boot partitions as well as others) connected to an Adaptec PCI-X SCSI-320 RAID controller.

Here's the history of this problem:

(1): Roughly two weeks ago, I discovered during boot time that the Marvell BIOS reported that it could no longer find the RAID 0 array definition for [bETA].

(2): All SMART tests and diagnostics on all hard drives, including [bETA] and [DELTA], passed perfectly. There was absolutely no indication of damage other than the missing array definition. Just as a safety precaution, I unplugged the two RAIDed drives from the Marvell to avoid any further problems with them. Note that the non-RAID [DELTA] drive connected to the Marvell was fine and all partitions on it were perfect, so I left that one plugged in.

(3): A few days later, after booting up I discovered that all the partitions on that non-RAID drive connected to the Marvell ([DELTA]) were missing and that it was marked "unallocated"!

(4): I then connected [DELTA] to one of the SATA ports on the ICH7R (since I assumed the Marvell had failed), then ran Disk Director 11 to recover the partitions, which worked fine. I left it on the ICH7R (there were now no drives still connected to the Marvell controller).

(5): A day or so later, I found all the partitions gone on one of the other drives on the ICH7R!

(6): Disk Director 11 recovered all those partitions, too.

(7): Later, all the partitions on the drive from step (4) disappeared again!

(8): I then recovered those partitions yet again!

(9): I just booted up to discover that all the partitions on yet a third drive connected to the ICH7R were gone now! (DD 11 recovered them, too).

The only drives that have never lost partitions are the 2 SCSI drives connected to the Adaptec! For what it's worth, I just finished running Kaspersky 2010 with maximum settings, but it didn't find any malware at all.

What on earth is going on?!? PLEASE HELP!

[Boris]

Sounds like some glitch with the Marvell controller ?

Or - just a long shot ??? but what PSU do you have powering all this ?

[Amby19]

Sounds like some glitch with the Marvell controller ?

Or - just a long shot ??? but what PSU do you have powering all this ?

Thanks for your reply, Boris!

I also was almost certain that the Marvell had blown. But then the same thing started happening with the drives connected to the on-board ICH7R, too!

My PSU is an expensive Antec 850 watt model. I've never heard any fans slow down or speed up, if that matters. I used SiSoft Sandra to monitor the voltages and fan speeds and temperature for about 20 minutes, but there was no significant change in these values over time.

Here's the report:

Board Temperature : 47.0°C (Min 47.0°C; Avg 47.0°C; Max 47.0°C)

CPU 1 Temperature : 34.0°C (Min 32.0°C; Avg 37.7°C; Max 47.0°C)

CPU 2/Aux Temperature : 45.5°C (Min 0.5°C; Avg 33.2°C; Max 53.5°C)

CPU 1 Fan : 1004rpm (Min 1004rpm; Avg 1056rpm; Max 1082rpm)

CPU 1 DC Line : 1.19V (Min 1.19V; Avg 1.24V; Max 1.31V)

CPU 2/Aux DC Line : 0.89V (Min 0.89V; Avg 0.89V; Max 0.89V)

+3.3V DC Line : 3.27V (Min 3.27V; Avg 3.27V; Max 3.27V)

+5V DC Line : 6.63V (Min 6.60V; Avg 6.62V; Max 6.63V)

+12V DC Line : 12.84V (Min 12.84V; Avg 12.84V; Max 12.84V)

Standby DC Line : 3.29V (Min 3.29V; Avg 3.29V; Max 3.29V)

Battery DC Line : 3.29V (Min 3.29V; Avg 3.29V; Max 3.29V)

CPU 1 Core Power : 51W (Min 51W; Avg 55W; Max 61W)

GPU 1 Temperature : 55.0°C (Min 54.0°C; Avg 54.7°C; Max 55.0°C)

GPU 1 Fan : 2100rpm (Min 2100rpm; Avg 2100rpm; Max 2100rpm)

The mobo and GPU 1 temps are high, so I just shut that system off.

Can you think of any way to be certain if the PSU is the problem? Any other suggestions?

[Boris]

That Antec should be ample :) your voltages look fine to me - I was just checking in case you had a generic 500w PSU powering the rig.

Other than swapping it out for an equivalent unit (which I don't suggest unless you already have one or can borrow one) there is nothing I can think of.

Your board should cope with the eight drives you have - it is well within the spec.

As both controllers seem to be affected it is looking more like degradation or a fault with the mobo ? How old is it ?

[Amby19]

That Antec should be ample :) your voltages look fine to me - I was just checking in case you had a generic 500w PSU powering the rig.

Yeah, I did the wattage calculations with all those drives in mind as well as the (now) two PCI-X HBAs and knew I had to buy a hefty PSU. And I've had such a good experience with Antec (they've really gone the extra mile for me in the past!), they were the natural choice. The reason there are now two PCI-X BHAs on that system is because I bought a Sil 31240-based 4-port SATA-II RAID card as a substitute for the on-board Marvell RAID controller.

I took Asus' advice and set the hardware jumper for 100 MHz operation on both PCI-X boards rather than 133 MHz. Considering the flakiness, that seems prudent.

Other than swapping it out for an equivalent unit (which I don't suggest unless you already have one or can borrow one) there is nothing I can think of.

Your board should cope with the eight drives you have - it is well within the spec.

As both controllers seem to be affected it is looking more like degradation or a fault with the mobo ? How old is it ?

The mobo's about 4 years old. That's not bad, is it?

Any other ideas? I'm completely stumped! I can't tell you how grateful I am for your replies!