[CentOS] USB key problems on enterprise systems

Sat Mar 4 02:32:55 UTC 2006
Collins Richey <crichey at gmail.com>

Just came up with some interesting (read: frustrating) problems on
RHEL3 and RHEL4 today at work, and I've confirmed one of the problems
on my up-to-date CentOS4 system at home. I was forced into releasing
some RHEL3 systems to replace our aging RH9 systems without time for
adequate testing, and that leads to problems like these.

On our legacy RH9 systems, mounting a USB key is no problem, but I
can't get it to work on either RHEL3-U6 or RHEL4-U2 (= CentOS4). I'm
curious if any CentOS3 or CentOS4 users have any suggestions.

Here are the results from plugging in a USB Key and trying to mount it:

1. On RHEL3-U6, some systems get the following, and there is a workaround:

Message: USB device not accepting new address
Workaround: /sbin/modprobe -r ehci-hcd

Message: READ CAPACITY failed
...
Unable to read partition table

Workareound: none that I am aware of. The above workaround has no effect.

3. RHEL4-U2 and CentOS4 get the following:

No possibility to mount. As soon as you plug  in the USB key, the
system goes into a SCSI error loop until you unplug it.

I've googled extensively for an answer, found a few hits for the RHEL3
problem, but no useful solutions.

The RHEL4 problem is apparrently a 2.6 kernel bug. Doe anyone know of
a packaged kernel solution for this? I'm using the CentOS unsupported
kernel at home, but same problem as in the stock RHEL4 kernel. Have RH
possibly fixed this in the U3 beta?

In case you're wondering about the key(s), the mount works flawlessly
including KDE automounting the key on my Kubuntu Dapper development
system with kernel 2.6.15-16-386 #1 PREEMPT.

TIA,

--
Collins Richey
      The agnostic dyslexic insomniac lies awake wondering if there is a dog.