I administer a remote server via SSH that runs CentOS 5.5. I have been unsuccessful in all my attempts to write to two different external USB hard drives with a single ext3 partition when logged in as root.

When attempting to create a "test" directory I get one of two messages:


Both drives *appear* to have filesystem issues. When I run an fsck on either drive, I get:

Keep in mind this is a newly-formatted, empty drive.

Not putting stock in the odds that I've had two hard drives (different sizes and brands) with the exact same hardware problem, I'm going to assume this is a software issue, although maybe it isn't. Hence, my post in "Linux - General". I've heard talk elsewhere of controller (chipset) issues coming into play. Is this valid?

Okay, here's the information you'll need to make a diagnosis....

Here's the output of a "df -h" command:


Here's the contents of my /etc/fstab:


Here's the output of "cat /etc/mtab":


Here's the output of a mount command:


Here's the output of fdisk on the device in question:


I've got someone with access to the box if necessary. But it might take days to implement solutions since this isn't his full-time job. Remote solutions are, therefore, preferable.

Many thanks!

I should also add that I've disabled SELinux and that has no effect.

Since you have two devices with the same error I don't think that your USB-disks themselves are faulty, I think that an other device is faulty. I would check at least this:
1. Are they connected with the same cable? If so, change it and give it another try.
2. Are the drives always connected to the same USB-port? If so change it and give it another try.
3. Check the drives on a different machine, to be sure that they are not the culprit.
4. Try a different device on this machine.

Answers:
1. Only one drive is in use at this time. Unsure if a different cable was used.

2. Uncertain if a different port was used each time.

3. They were previously in use as Windows drives. They've been reformatted (using CentOS) and only now are they having issues.

4. I have a 3rd drive but do you really think it's 2 bad drives?

As I said, I'm 400 miles away from this server. I can have someone attempt the above solutions but it will take time.

Any software suggestions to try in the meantime?

Thanks so much for your help.

I am asking this only because it has not been directly addressed and apologize if I've missed something obvious.

When the drives were partitioned and a file system type selected, was mkfs run on them to create the file system?

@frankbell:

I'm on my 2nd hard drive. The first hard drive I partitioned myself using fdisk utility. No go. The 2nd time around a colleague partitioned it and I'm not sure exactly how he went about it. Is there some kind of advantage to using mkfs? Do you have a command suggestion? If so, I can pretty easily repartition the drive.

For 1+2, you have to make that sure. For 3, frankbell's question is a good one. For 4, such coincidences happen very seldom, but it is the simplest way to use a known-good device on that machine to make sure that the drives are not the culprit.

Without using mkfs there will be no filesystem on the disks, if you didn't use a partitioning tool like GParted that formats automatically. The command should look like this:Assuming that sda1 is the partition on the external disk.

@TobiSGD

I apologize for the confusion. I thought frankbell was only asking about how it was partitioned, not how it was formatted. I missed that part. It is formatted ext3 and an "ls -la" command produces the following:

That directory listing is from root, by the way (the drive is mounted as /mnt/usb).

Full disclosure: I'm a web admin and not accustomed to formatting hard drives since I don't have access to any servers. Usually that is done by others. As is adding hard drives or other hardware. So please bear with me if I miss some obvious questions. In this particular instance, the client is running a backup server (the computer in question) in their office. There is a tech who moonlights for them who can swap hardware when he gets a chance but otherwise, this is all on me.

Okay, I've asked the tech who is local to the server to try TobiSGD's suggestions. I'll report back with results once I have them.

Okay, here's the report from my man on the ground:

So it appears he can write to the drive in FAT but not EXT3.

Thoughts?

A quick update from my tech who was at the server this morning.... Apparently, if you disable the ehci_hcd module, which disables USB 2 in favor of 1.1, the drive is writeable under ext3. This suggests that something like the bug involving nForce2 chipsets (nVidia motherboard) applies. However, I'm awaiting confirmation of which motherboard chipset the server uses.

Because this is a backup server that transfers multiple gigabytes of data in a single script execution I can't fall back on USB 1.1 permanently. The workaround seems limited to the current session only which is also unacceptable. Perhaps I'm misunderstanding the workaround or someone knows of a more elegant, permanent software solution?

Thanks for all the help thus far!