Coda File System

Re: problem with 2 servers

From: <jaharkes_at_cs.cmu.edu>
Date: Wed, 23 Sep 1998 21:23:28 -0400
> Hi !
> 
> We set up 2 coda servers. One scm and one normal.
> Then we created 3 volumes. So far it worked ok. All volumes were replicated
> on the server.
> But it is not possible to clog on the non scm maschine although venus is
> running...
> Does anyone know what we can do ???

Hi Thomas,

Obtaining an authentication token for coda is a two step process, where clog
first contacts an auth2 server on an arbitrary server machine. (I believe it 
picks a name from /usr/coda/etc/vstab) and gets an encrypted token.

Possible problems here are:
  - auth2 daemon is not running.
  - there is no up-to-date user information.
  - auth2 daemon does not have up-to-date user information because
    updateclnt/updatesrv are not running.

Then clog contacts venus and passes the token it received. Venus then uses
this token to set up a secure connection with the server.

Possible problems why the server might reject a token:
  - clock skew between auth2-server/coda-server
  - corrupt/bad token (is that auth2.tk shorter than 8 characters bug still 
	there?)
  - auth2-server didn't have up-to-date user information, see above.


The initial places to look for more information are:

 check if auth2/rpc2portmapper/updatesrv/updateclnt are all running.
 check /vice/auth2/AuthLog
 check /vice/srv/UpdateClntLog

on both servers for error messages.

succes,
	Jan
Received on 1998-09-23 21:27:25