eSATA harddisk corrupting data

tomblue · Posted: Tue Feb 17, 2009 4:29 pm Post subject: eSATA harddisk corrupting data

Hi all,

I've got a Seagate ST3160023AS harddisk (160GB) in an Enermax JAZZ HDD enclosure attached to a MSI P7N mainboard via eSATA.

If I copy a file bigger than round about 700MB, the data (haven't figured out the exact threshold) is getting corrupted. MD5 sums of the original and the copy differ. Smaller files seem to be ok though.

I'm running Gentoo (kernel 2.6.25.9) as have tested with several live CDs (e.g. knoppix [kernel 2.6.28] or systemrescuecd [kernel 2.6.27]). They all show the same problem. The puzzling twist is: A parallel installed Vista64 does NOT show such odd problems!? That makes a hardware issue unlikely I suppose.

Has someone seen such an behaviour? Anyone an idea what might cause this?

Some more hardware details:

Intel Q9450@2,66GHz
MSI P7N Bios 1.3 (latest available at the moment)

eSATA controller

petlab · Posted: Wed Feb 18, 2009 7:04 pm Post subject:

Its possible you just have cable problems. I know that answer sucks, but I have read about and experienced random problems with sata cables in general. If you have another cable, try that.
_________________
Get Serious - Get JAWA CZ

tomblue · Posted: Wed Feb 18, 2009 7:18 pm Post subject:

Hi petlab,

I don't have a spare cable unfortunately. As the drive works fine with Windows Vista, using the very same cable, I would rule out hardware issues.

I was looking for new, longer cable anyway, but has not yet arrived. I'll definitely give it a go once I've got it in my hands.

Cheers

SysGhost · n00b Joined: 24 Apr 2007 Posts: 48 Location: Stockholm

Well. You have ruled out the hardware as a source of failure.

Kernel module for serial ATA is highly unlikely to fail.

So whats left?

The filesystem

Are you using NTFS ?

If so try another filesystem as I believe that the NTFS module included in the kernel have a somewhat "troubled" write support.
It is after all marked "EXPERIMENTAL".

tomblue · Posted: Thu Feb 19, 2009 6:59 pm Post subject:

I received the new eSATA cable today - no change in behaviour.

Filesystem could have been a problem, good point. I checked with several different ones though: NTFS, ext3 and jfs. Still the same problem... :cry:

SysGhost · n00b Joined: 24 Apr 2007 Posts: 48 Location: Stockholm

This is strange.

Works fine under Windows.
Hardware works fine.
Filesystem seems not to be the matter.

Conclusion: Kernel module.

Are you using the general AHCI SATA kernel module or a hardware-specific kernel module?

I would suggest to use AHCI as far as you can if the hardware supports AHCI mode.
I believe the hardware-specific modules for many controller chips/cards are not working too well with the hardware since some modules are related or even themselves reverse-engineered and thus "experimental".

tomblue · Posted: Sat Feb 21, 2009 2:40 pm Post subject:

The AHCI module is supposed to power the eSATA drive. I've enabled AHCI for the controller in BIOS.
The NV (Nvidia) module supports the internal SATA interfaces and currently manages to work fine with 3 attached hard disks.

SysGhost · n00b Joined: 24 Apr 2007 Posts: 48 Location: Stockholm

If possible, try using AHCI on all devices.
I'm not sure if the sata_nv works well with AHCI, but once upon a time I tried the nv module took over everything leaving the AHCI without "work".
Later on I disabled the nv module and I was lucky enough to have hardware that was fully compatible with AHCI. It solved some trouble for me back then.

If your onboard nividia sata is AHCI-compatible, consider it.

tomblue · Posted: Mon Feb 23, 2009 7:15 am Post subject:

I'll give it a go and post an update asap.

Thanks for the help so far!