[OpenAFS] openafs-1.4.2 RHEL RPM package installs nonempty SuidCells and mangles CellServDB

ted tcreedon tcreedon@easystreet.com
Fri, 17 Nov 2006 19:40:20 -0800


SuSE uses /etc/sysconfig/afs-server and /etc/init.d/afs-[server,client]

Can be cribbed.

tedc

-----Original Message-----
From: openafs-info-admin@openafs.org [mailto:openafs-info-admin@openafs.org]
On Behalf Of Carson Gaspar
Sent: Friday, November 17, 2006 7:28 PM
To: openafs-info@openafs.org
Subject: Re: [OpenAFS] openafs-1.4.2 RHEL RPM package installs nonempty
SuidCells and mangles CellServDB

Derek Atkins wrote:
> Derrick J Brashear <shadow@dementia.org> writes:
> 
>> However, the thing that may make sense is a package which does not munge 
>> configuration which conflicts with the regular package but can be 
>> installed in place of it, to address this issue. I'm unsure how we could 
>> handle documenting it, as I'm sure everyone who doesn't bother to read 
>> instructions will try to install both, get a conflict, and be confused.
> 
> I figure that sites that care THAT much about custom configuration
> can just repackage the RPMs themselves.  It's not that hard to
> override the scripts.

We are trying to get away from our custom init scripts, as they are a 
significant maintenance burden, especially when packages change options. 
We'd like to benefit from improvements in the packages (such as letting 
the openafs client auto-tune instead of hard-coding performance options, 
etc.). In my ideal world, the script would:

- Source /etc/sysconfig/$something (openafs/openafs-client/...)
- Allow the DB munging to be disabled by some variable set in the file above
- Have _extremely_ robust error checking in the DB generation code, so 
it doesn't leave mangled files if the filesystem is full.
- Manage the cache location and size (if set in the sysconfig file), so 
all the variables are in /etc/sysconfig instead of squirreled away in 
/usr/vice.

I'd be happy to work on the changes if there was interest in 
incorporating them (our code already does a lot of this). If not, I'll 
just keep maintaining our local init script.

-- 
Carson
_______________________________________________
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info