First attempt with arla-0.6

[LeBlanc@mcc.ac.uk]

I wrote:
> avl 13:40:59 > ./startarla
> avl 13:41:17 > lsmod
> Module         Pages    Used by
> xfs                6            2
> ...
> avl 13:41:25 > mount
> ...
> arla on /afs type xfs (rw)
> avl 13:41:28 > ls /afs/
> ls: /afs/: Not a directory
> avl 13:41:40 > file /afs

"A.J.Martin" <A.J.Martin at qmw.ac.uk> wrote:

>Are you using a glibc2 system?  

Yes

Assar Westerlund <assar at sics.se> wrote:

>Per default that output would go to syslog.  I suggest running `arlad
>-n --debug=all' which will not fork and send the debug output to
>stderr.

Aha!  This has now produced debug information.  But first, I wrote:

> The kthkrb4 stuff is installed, but it is the December version, and
> does not recognise arla as AFS, so that, for example, pagsh doesn't
> actually get a PAG.

Assar Westerlund <assar at sics.se> wrote:

>The easiest way of fixing that should be copying /usr/arla/etc/* to
>/usr/vice/etc/.

This did not in fact make any difference; the kthkrb stuff still
doesn't recognise arla as AFS.

Assar Westerlund <assar at sics.se> wrote:

>This sound like what would happen if you have mounted the xfs
>filesystem but don't have any running arlad, in that case xfs would
>create a fake root.  Can you check that you actually have a running
>arlad and see from the debug output of that if it gets any message
>when you try to access /afs ?

Yes, arlad is running.  The first time I ran it with '-n --debug=all'
I noticed the error message

arlad: fprio_readin: cant open file /usr/arla/cache/fprio: Operation not permitted

so I tried creating this file using touch, and afterwards did not have
this error message.  This is the debug output from arlad:

arlad: read_conffile: /usr/arla/etc/arla.conf
arlad: Arlad booting sequence:
arlad: connected mode: connected
arlad: initports
arlad: conn_init numconns = 100
arlad: initconncache
arlad: cellcache
arlad: fprio
arlad: volcache numvols = 100
arlad: recovered 1 entries to volcache
arlad: rx
arlad: using rxkad level auth
arlad: credcache numcreds = 100
arlad: fcache low_vnodes = 3000, high_vnodes = 4000low_bytes = 15000000, high_bytes = 20000000
arlad: get_info: we asked for 1835374534 and got root.afs
arlad: recovered 1 entries to fcache
arlad: cmcb
arlad: cm
arlad: arla init done.
arlad: running cleaner: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: cleaner done: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: Arla: selecting on fd: 6
arlad: running cleaner: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: cleaner done: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: running cleaner: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: cleaner done: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: running cleaner: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: cleaner done: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: running cleaner: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: cleaner done: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: running cleaner: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: cleaner done: 1 (3000-4000) files, 0 (15000000-20000000) bytes
arlad: fatal signal received
arlad: storing state
arlad: wrote 1 entries to fcache
arlad: wrote 1 entries to volcache

The 'running cleaner' message appears to occur whenever I do 'ls -l /afs'.
The 'fatal signal' received was the Ctrl-C with which I killed the
debugging arlad.

>The debug output from xfs is probably also useful.

How do you get debugging output from xfs?  I tried to read the source,
but I couldn't make this out.

     -- Owen
     LeBlanc at mcc.ac.uk