Coda File System

Re: Crashed server fails to restart

From: <>
Date: Fri, 14 Sep 2007 14:16:56 +0200
On Fri, Sep 14, 2007 at 12:56:14PM +0200, Martin van Es wrote:
> My (single) server crashed the other day, and now my volumes are in kind of a 
> messed up state. The server quits with the following message (in SrvErr log):

> The problem is, this is kind of a catch-22, because even if I'd wanted to 
> purge all volumes and start from scratch I would need a connection with the 
> vice daemon, but that refuses because of errors in the volumes (I guess). So 

Hi Martin,

in fact, you can do things with the server data even if the server
process can not start. Look (in the archives?) for examples of "norton" usage.

> out how useable coda is. Like I said before on this list I do like it a lot, 
> however at the moment I would hesitate to implement this in even a small 
> production environment with 'shortcomings' like this.

To feel safer you can begin with setting up backups and checking
that you can restore arbitrary or all of the files even in the absence
of a working Coda realm.

Some important things for using Coda in production are:

- to find out what kind of activities is incompatible with Coda
  (both because of fundamental limitations like write-sharing
   and also to avoid touching yet-buggy codepaths)
- tolerant users who also can learn "donts" for the above reasons

Then you will be fine^W^W^W^W^WYMMV :-)

Received on 2007-09-14 08:18:46