Aaron Lehmann <[EMAIL PROTECTED]> wrote:
>
> Hi.
>
> I've just set up a coda server for my home on a i386 Linux 2.2 machine.
> When I run codasrv.init start, codasrv is started and starts using 100% of
> available CPU. The howto says that codasrv does a "fsck-type" check on
> boot, so I waited for awhile to see if it would finish. After three
> hours, it was still using 100% of cpu (BTW, 15% user and 75% system). It
> never gets to the point where it has completed booting and other clients
> can connect.
>
> Coda looks cool, I'm looking to use it instead of NFS for my MP3's and
> sharing home directories accross machines. I'm not really in the mood for
> repartitioning the drive on my server though. What is the effect on
> performance of using files and directories on the normal ext2 filesystem?
> I'm not planning to do any signifigantly intensive tasks, mostly streaming
> of MP3's and accessing files (I hope to have everything cached, at least
> on my laptop so I can take advantage of disconnected operation). The LAN
> is 10mb and there will be a max on one simaltanious client actually
> accessing files.
>
> Server version is 5.2.7.
>
> Here is as far as the log files get:
>
> 18:04:51 New SrvLog started at Sun Jun 27 18:04:51 1999
>
> 18:04:51 Resource limit on data size are set to 2147483647
>
> 18:04:51 Server etext 0x80fe4a2, edata 0x8139ca8
> 18:04:51 RvmType is Rvm
> 18:04:51 Main process doing a LWP_Init()
> 18:04:51 Main thread just did a RVM_SET_THREAD_DATA
>
> 18:04:51 Setting Rvm Truncate threshhold to 5.
>
>
> Date: Sun 06/27/99
>
> 18:11:43 Starting new SrvLog file
Just a wild guess. Is it because your codasrv had swap to use another
log file (as indicate by the last line of your log) yet you were still
looking at the old log file ? You can look into your /vice/srv, the
log files are called SrvLog and SrvLog-{1,2,3,...}.
You can also know if your codasrv is up and running by seeing (in the
most current log file, i.e. /vice/srv/SrvLog) if there is a line like
14:03:34 File Server started Tue Jun 22 14:03:35 1999
Also, from a client, you can use "cmon" to know if the server
is working properly. I usually use the following alias, which
starts a small xterm and then fires up cmon to look at my two
Coda servers (xylophone and violin).
[clement@aria clement]$ alias xcm # I am using bash
alias xcm='xterm -title " CMON" +sb -fn 6x10 -geometry 34x25+0-25 -e cmon xylophone:64
violin:64 &'
And the xterm will show something like
xylophone violin
TIM
mon 12.0 days 12.0 days
prob 13:36:56 13:36:56
succ 13:36:56 13:36:56
up 6.0 days 6.0 days
bind 2 2
CPU
sys 0 0
user 0 0
util 0 0
RPC
conn 21 20
wkst 3 3
call 1 1
pki 1 1
pko 1 1
byi 640 640
byo 640 640
DSK %
max1 vicepa:46 vicepa:76
max2 *** ***
max3 *** ***
which indicates that my two servers are running fine (for 6 days).
BTW, are you really using an 80386, or you are using some more
faster CPUs (but compatible to "i386") ?
(As a last resort, you may use gdb to attach to the codasrv
process and see at what point the codasrv may be infinitely looping.
Of course, this is much more involved.)
-- Clement
======================================================================
Yui-wah LEE (Clement) Tel: (852)-26098412
Department of Computer Science and Engineering, Fax: (852)-26035024
The Chinese University of Hong Kong Email: [EMAIL PROTECTED]
======================================================================
