Coda File System

Re: Installing Coda in Linux (Debian)

From: Paulo Henrique Baptista de Oliveira <baptista_at_cos.ufrj.br>
Date: Mon, 6 Sep 1999 15:43:41 +0000
	Hi Jan,
Quoting Jan Harkes (jaharkes_at_cs.cmu.edu):
> On Mon, Sep 06, 1999 at 01:03:54PM +0000, Paulo Henrique Baptista de Oliveira wrote:
> > My two questions are what is the cache size? and how do I know/set
> > "adminuser" passowrd?
> 
> Cachesize is how much diskspace the client should use on the local disk.
> In the order of 20000 - 100000 diskblocks (20MB - 100MB) is probably a
> good choice. Oh, and this is only an `indication', Coda only detects the
> filesize when a file is closed, so newly created files will overflow
> the cache. Also as a safety net, when a client is logging operations
> (working disconnected or write-disconnected), it dumps modified files in
> a tar compatible archive on disk. The admin password is probably still
> `changeme', you can change it with 'au -h <server> cp', or cpasswd.
> 
> Maybe it is easiest to verify the client setup by connecting to the
> testserver.
> 
>     # venus-setup testserver.coda.cs.cmu.edu 40000
>     # venus
> 
	I tried this and the answer was:

pixinguinha:/usr/local/home/baptista# venus-setup testserver.coda.cs.cmu.edu
40000
/etc/services already has new services registered! Good.
pixinguinha:/usr/local/home/baptista# venus
Date: Mon 09/06/99
15:27:09 /usr/coda/LOG setup for size 0x106c64
15:27:10 /usr/coda/DATA initialized at size 0x41b190
15:27:10 brain-wiping recoverable store
15:27:10 loading recoverable store
15:27:13 starting VSGDB scan
15:27:13        0 vsg entries in table
15:27:13        0 vsg entries on free-list
15:27:13 starting VDB scan
15:27:13        1 vol entries in table (0 MLEs)
15:27:13        0 vol entries on free-list (0 MLEs)
15:27:13 starting FSDB scan (1666, 40000) (25, 75, 4)
15:27:13        0 cache files in table (0 blocks)
15:27:14        1666 cache files on free-list
15:27:14 starting HDB scan
15:27:14        0 hdb entries in table
15:27:14        0 hdb entries on free-list
15:27:14 Initial LRDB allocation
15:27:14 Getting Root Volume information...
15:27:29 GetRootVolume: can't get root volume name!
15:27:59 GetRootVolume: can't get root volume name!
15:28:29 GetRootVolume: can't get root volume name!
	.
	.
	.
15:29:29 GetRootVolume: can't get root volume name!
15:30:32 RecovTerminate: clean shutdown
	And I typed Ctrl+C

Something is wrong! Or not?
	Thanks for any help.
	If I install Coda sucessful I will try in my lab with 20 machines
running Debian GNU Linux! :)


> And when it says /coda is mounted, in another window
> 
>     $ ls /coda
> 
> 
> To connect to you own server, kill the running venus, unmount /coda,
> reconfigure venus for you own server, and start:
> 
>     # vutil shutdown  (or kill -TERM `pidof venus`)
>     # umount /coda
>     # venus-setup <server name> 40000
>     # venus
> 
> 
> If you fail to connect to your own server, check that all the necessary
> servers are running,
> 
>     # ps auxwww | grep rpc2portmap	# these 3 are necessary to
>     # ps auxwww | grep updateclnt	# create new volumes
>     # ps auxwww | grep updatesrv	
>     # ps auxwww | grep auth2		# authentication (logging in)
>     # ps auxwww | grep codasrv		# the actual file server
> 
> Check the server logs:
>     /vice/srv/SrvLog
>     /vice/srv/SrvErr
> 
> Then check if it has a rootvolume
>     cat /vice/db/ROOTVOLUME		# should have a single volume name
>     cat /vice/vol/VRList		# should have this volume name too
>     cat /vice/vol/VolumeList		# here is it volumename.0
> 
> 
> > After setting CODA I will do some performance tests against NFS.
> 
> My own experience is, slightly slower in connected mode, a lot faster
> (and more unstable) when write-disconnected. A lot of the w/d
> instabilities did get fixed in 5.3.1.
> 
> > Please help a student in a hurry. :)
> > 
> > Thanks,         Paulo Henrique
> 
> Jan
Received on 1999-09-06 14:33:06