I can usually just plug in a USB device into a slot and type mount /dev/sda1 and everything is fine. But *sometimes* when I type mount /dev/sda1 I get:

I ticks me off. How do I fix it?--I reboot the comp and then type mount /dev/sda1 and it just works. Why do I have to reboot the comp to do this?

What's the output of dmesg?

Are you sure it's always at /dev/sda1? It you have other devices (scanners etc show up as SCSI devices), it may be something different.
It can also take a couple of seconds for the device to be "ready". Occasionally, I get a similar error, waiting a few seconds until the device is ready to talk usually suffices.

If you're using 2.6, you can use udev to assign a permanent name to the device. I haven't got around to setting up udev yet, so I can't offer any details.

In any case, you can always label the filesystem and use that instead of /dev/whatever. With ext2/3, you can use e2label:

The label name can be whatever you want. Then modify /etc/fstab, replacing /dev/sda1 with LABEL=<label name>.

Instead of using the device name or label with mount, it's better/easier to use the mount point, e.g.

or whatever.

Ok, I will try using the mount point instead of the device name. Its just so odd how sometimes it will work and sometimes it wont. Usually if I mount and then umount the device and then try to mount it again without rebooting it refuses. I got it in /etc/fstab already and I am using a 2.4 kern.

That certainly sounds strange. What happens if you unplug it and plug it back in? What does dmesg report?

You'd also get a "not a valid block device" error if one of the necessary modules was being unloaded for some reason (though a simple umount should not cause modules to be unloaded). You need usb-storage, scsi support of course, and sd_mod if it's not built into the kernel. A while back, I was using a kernel with scsi support configured as modules and sometimes sd_mod wasn't being modprobed but I can't remeber why. I currently have scsi support built into the kernel, with usb-storage as a module. Sometimes, a 'depmod -a' can fix that, though that should be run automatically by the pkg manager when you (re)install a kernel.

Another thing that can cause problems is the LUN (logical unit number) scan. I had an external SCSI drive years ago (like a zip, but a different make of course) that screwed up if I had the 'Probe all LUNs on each SCSI device' option set, though modern devices shouldn't have that problem.

On the other hand, if you have one of these multi-card readers (cf, sd, etc), you generally need that option turned ON for it to work. I have a multi-card reader which I seem to recall behaved similarly to what you describe until I turned on the LUN probe option, but a USB drive *should* just be a single device so the LUN option shouldn't matter. Unless the've set it up as mutiple logical devices for some reason.

I have that problem sometimes ... What I do to fix it is the follwing:
Open up a terminal
log on as root
run : fidsk -l /dev/sda*

What this does is list everything connected to a /dev/sdaX, the computer will then regiester that the drive is there and you can mount it.. saves you a reboot.