Dear lazyweb,
I’m currently struggling with an utterly annoying problem that’s preventing me to do photo processing. Since I’m a photographer for some of my time, that’s really unfortunate.
Here’s the deal: I have a USB Hard Disk, a Western Digital Passport Elite (320Gb) and no matter what filesystem I put on it, after a day or two of working, it becomes corrupted.
I’ve run a bad blocks test (no problems), I make sure I unmount it cleanly each and every time and I’ve been extra careful with the drive.
I don’t recall this happening a few months ago (on Ubuntu Hardy), but recently it has started happening repeatedly (now running Intrepid).
I have no clue what causes this, but when it happens, I usually see something like this in dmesg:
[108400.182508] hub 4-0:1.0: over-current change on port 3
[108400.288094] hub 2-0:1.0: over-current change on port 1
[108400.296192] usb 4-3: reset high speed USB device using ehci_hcd and address 33
[108446.992602] hub 4-0:1.0: over-current change on port 3
[108447.096102] hub 2-0:1.0: over-current change on port 1
[108447.104075] usb 4-3: reset high speed USB device using ehci_hcd and address 33
[108828.704908] EXT3-fs error (device sdb1): htree_dirblock_to_tree: bad entry in directory #14770193: rec_len is smaller than minimal - offset=0, inode=0, rec_len=0, name_len=0
[108828.707727] EXT3-fs error (device sdb1): htree_dirblock_to_tree: bad entry in directory #14770199: rec_len is smaller than minimal - offset=0, inode=0, rec_len=0, name_len=0
Does anyone have a clue what might be affecting me? Is this my hardware, or is something wrong with my software?
Western Digital Passport Elite (320Gb)
I’ve seen similar issues with USB drives in the past, I would check to see if there is a newer firmware for the USB-IDE controller as I seem to remember that helped the issue.
If you can afford to take the disk out of the casing and put directly in a laptop, you can boot with a live CD and try the smartmontools. With these tools, you can view the S.M.A.R.T. variables, see whether something really bad happened to the device.
I do not think that smartmontools will work directly through the USB interface.
While not helpful in solving the issue, I am experiencing the same issue with my 700GB WD external drive. Fsck will report tons and tons of errors but running spinrite (for 30 long miserable hours) on the drive to detect bad blocks reveals nothing.
This was on Intrepid, I am currently back on Fedora 10 and I will see if the issue crops up here as well.
Peter: Am running the most recent firmware, but thanks for point out this could be a problem.
Simos: There’s no way to break this thing open, but apparently Western Digital has a (windows-only) diagnostic tool to read those values. Once I get hold of a windows laptop, I’m going to do the thourough diagnosis.
David: Cool (well, not really), let me know what happened, it might be interesting if it turns out Fedora doesn’t have it (which means we’re seeing a software bug that can be fixed).
If you have an external power supply you can plug into the drive, I’d try that.
The photo makes it look like you might only have it plugged into the USB port and it might be drawing more current than the port can support.
Is the drive powered via USB or an external brick? If it’s an external brick, it may be going bad. Symptoms in such cases usually involve random resets and issues followed by long periods of workingness. Additionally, you might try a different USB cable (possibly thicker – lots of stuff ships with horrible out of spec cables), and verify physical integrity of both USB connectors.
“over-current change” means that your USB port cannot sustain the disk energy requirements. I think USB ports have a maximum 500 mA ouptut (and some laptops even less). I have a USB 3G modem with a Y-cable that “gets” energy from two USB ports. My Dell laptop has 4 USB ports and the modem only works with 2 of them – the other 2 cannot give enough current.
And if you have more USB devices, they all add up. Try removing all other USB devices or using a external powered USB hub between you laptop/PC and the disk.
I’ve seen perfectly good USB HDDs have problems on on certain machines, but they work fine on others. The errors you show are similar to ones I’ve seen. I’ve not yet had a chance to determine if it’s a software thing or a hardware thing, but I can tell you that it’s not specific to Intrepid. My most recent “problem machines” were Dell Poweredge 1435 (I think) running Debian Etch. They reported bad sectors and generally wigged out with a Seagate 2.5″ HDD, which works fine on every other machine I’ve used it on (most of them running Ubuntu Hardy). My theory at this point is that these drives tax the power supplied to the USB ports (even with the Y-USB cable) and either the hardware or the driver has a hard time regulating the power well enough to keep everyone happy.
These Passport drives draw power from the USB port. They won’t even spin up if there’s not enough power available on the given port. It sounds like your drive is actually drawing more power than it says it needs, hence the “over current” error? If it doesn’t get the power it needs, it is not hard to imagine how data corruption could happen.
Try running it through a powered hub, and see if that solves your problem. I have a powered hub precisely for this reason, as most USB ports on most of my machines (desktop, laptop, N810) don’t have the power needed to run a Passport drive.
From the dmesg output it looks like your drive is drawing too much power, maybe that is causing the corruption?
Some 2.5″ drives need more power than the USB spec says a host must supply and some hosts don’t provide the minimum power the spec says must be supplied anyway.
I use an external 2.5″ drive for tunes and backups, my X61 notebook’s USB ports provide enough power for it but my workstation at the office does not – I get spurious disconnects unless I use the supplied y-cable and plug it into two ports.
/Mike
Try also replacing the cable.
You could make sure the partition is the right size… if the partition is slightly larger than the size of the drive, you’ll see weirdness as soon as any data starts getting placed beyond the edge of the real drive.
fdisk should show if this is the case…
How is the HD powered? 2.5″ drives often draw too much power from the USB port (especially with laptops) leading to corruption. 3.5″ drives DEFINITELY do.
So, try powering the drive externally. Or, at the very least, plug it into a powered hub with a good, stiff 1A power supply.
Could this “over-current” thing mean your USB port isn’t strong enough to power the beast correctly?
I was having similar problems when I was connecting an external HD case with “regular” SATA (not eSATA). (Plain SATA is meant to be used internally only, but HD case manufacturers offered it nonetheless before there was eSATA.)
Since it was a data transport problem for me, it could well be the case for you even though you are using USB.
It sounds silly but I had similar issues with the Studio version of this HDD, and it was a dodgy cable causing all the problems. Have you tried using another USB cable yet?
I had the same problem with a LACIE 500 GB. I don’t think there’s a way out. If you search the Ubuntu forums, there’s an open bug about it. I eventually gave up and sold the HD. I think it happens only starting from a certain size on.
No problems with my WD Passport 120GB.
I’ve had similar problems some time ago with an IDE disk in an removable enclosure. Filesystem got silently corrupted every once in a while and no diagnostics (SMART, manufacturer’s utilities) ever reported any problems. The problem disappeared when I changed the enclosure, so it was probably because of some bad connections on the IDE connector.
If your USB enclosure contains a 2.5″ IDE disk, it may be that it slipped out of the connector. This doesn’t explain the overcurrent messages though.
I believe is a problem with the hardware itself – my father has one which he uses under Windows and the same issue occurs..
Hi,
I have a 250GB WD Passport, and it also gave me all sorts of weird problems until it stopped being recognized at all, either by Linux (Fedora 9) or Windows. Before I threw it away, I reminded that WD says that “an optional cable is available for the few computers that limit power from the USB port.”
I decided to try this out. It was not easy to find such optional USB cable here in Brazil, so I went out and bought an external case for 2.5″ SATA drives which is powered by a separate USB cable (this is important: you’ll find tons of cases which are *not* powered by a separate cable, which will not help you at all). I transferred the drive to the other case, and now it works like a charm, never had any other problem whatsoever.
(BTW: opening the WD Passport case is a real PITA)
Further experiments showed that 160GB Passport drives (which are not manufactured anymore) were the only ones that worked flawlessly with only one cable (my brother bought one). So, shame on WD: they should provide the extra cable along with 250GB+ drives. This whole “few computers might need an extra cable” speech is pure BS.
HTH
Make sure you plug the drive on the USB ports directly soldered to the mainboard.
The only thing I know about those messages:
[108400.182508] hub 4-0:1.0: over-current change on port 3
is that after having them for a while (and nobody knowing why they pop up) the usb ports on my laptop’s mainboard died …
I want to thank everybody for the abundant amount of comments. As it turns out, my hard disk was indeed underpowered (but not broken or anything, thank god!). I fixed the problem by attaching it using a Y cable, which draws the current from two USB ports. For safety, I’ve also added a powered hub in the chain. My desk looks like a mess, but it works, which means I can go back to developing photos.
Thanks all!
Replying to an awfully old post here, but what the hell. :-)
First of all: this issue is one I’ve seen a lot over the past few years. IIRC, I already filed some bugs about it against Ubuntu Warty’s. The problem seemed to disappear on some kernel updates, then reappear on others. Hence I always thought it was a kernel bug.
Secondly: there is a way to crack open this WD case. In fact it’s the cheapest way to extend laptop harddrives. After the holidays I’m gonna buy a 500GB WD, crack it open and put the harddrive in my MSI Wind. I guess you’ll find that interesting too. ;-)