Q. Are there any test WebDAV servers?
A. See these two websites.  (They are NOT DAV mount points.)
   - http://ocean.cse.ucsc.edu
   - http://test.webdav.org

Q. How many WedDAV server can I mount?
A. Maximum five. 

From: Jan Harkes <jaharkes@cs.cmu.edu>

     linux/include/linux/coda_psdev.h

    #define MAX_CODADEVS  5    /* how many do we allow */

Q. It looks like if minor number is grater than 5, Coda can't support the file.
 Am I right?

A. Correct, the array in the kernel doesn't fit any more than that.

Q. What happen, if we want to mount more than five servers?

A. You have a problem, of course you can change that define and recompile,
but that gives you a bunch of new problems.

- Whatever you set the new value to, you still cannot mount more devices
  than the value is. Each additional 'possible' device will use up
  kernel memory. And it can't be more than 255 anyways, minor number is
  only a byte. So it doesn't really solve your problem, it just delays it.

- Device major/minor numbers are scarce in the kernel and it won't take
  long before others are given a chunk of the Coda major space, so if
  you extend MAX_CODADEVS you would likely collide with other devices.
  As a result you could get more serious problems.

- You you need to convince everyone that uses your system to recompile
  their kernel with the new limit, both using up precious unswappable
  memory and risking device number collisions.

Q. If coda is compiled as a module, how can I insert module automatically?
A. You need to add an entry to /etc/modules.conf (or /etc/conf.modules on some systems)

    alias char-major-67  coda  # this one is needed for normal systems
    alias /dev/davfs*    coda  # this one is used by devfs


Q. I've got segment fault when I write something. For Example, 'cat > aa.c
' gives me a segment fault.

A. Bt Jan Harkes
That's probably because you're using ext3 which was afaik not in the
official kernel at all, RedHat decided to add it and use it as a default
for newly created filesystems. However ext3 and Coda didn't like each
other, or rather, ext3 didn't like the way Coda was directly accessing
the underlying files.

The initial patches only got merged in 2.4.10, which was a disaster
kernel in all other areas (but Coda worked well). By the time 2.4
stabilized again with the new VM, a few more bugs cropped up in the new
code which finally got fixed in 2.4.19. The only patches I have in my
queue right now are an extraneaous unlock_kernel in a failure path, and
a deadlock in readdir when you pass the fd of an actual directory
instead of a file with a bsd style directory layout.

Q. Which kernel do you prefer?
A. I would recommend to use linux kernel 2.4.19 or later version.