Coda File System

Re: Building Coda 6 on FreeBSD (was: Re: "Unknown error: 198" during re-integration)

From: Jan Harkes <jaharkes_at_cs.cmu.edu>
Date: Sun, 6 Jul 2003 16:10:16 -0400
On Sun, Jul 06, 2003 at 03:19:45AM +0200, Peter Sch?ller wrote:
> Ah, thanks!
> 
> About 6.x - is the vanilla source compilable on FreeBSD (4.8)? I am getting 
> errors pretty fast about redefinitions of types when it compiles 
> kerndep/pioctl.c (which I assume is needed for the client only?).

6.0.1 should be able to build a working server, but for the client you
probably would want to use what is currently in CVS.

> I understand from the previous thread that someone supposedly has 6.x, even 
> the client, working on FreeBSD 5.x, which would indicate 6.x is "supposed" to 
> work on FreeBSD. Any special procedure needed to compile it, beyond what is 
> in the cocs? Is the FreeBSD realms patch on the FTP site (for 4.8 and 
> current) for realms support for 5.x, or is it meant as a patch to make the 
> Coda 6 client work?

I'm not sure about any special procedures, the 'current' patches were
against CVS somewhere around the time that the diffs were made. And yes,
I know they are not unidiff and thus will probably apply incorrectly
when changes are made in the same files as the patch touches.

> I was going to stick to 5.x for now and wait for 6.x to make it into ports, 
> but if the re-intergration problem is fixed in 6.x I'd like to have a shot at 
> it.

You can use a 6.0.x server even if all your client are still running 5.3.x.
This gives both the reintegration and resolution fixes, and the clients
can live without realms, kernel patching and CVS versions just a bit longer.

> MIght have been it, not sure. I'll have a go next time. But the kernel support 
> is definitely killed because "df" yields complaints about missing upcalls 
> afterwards.

No, that actually means that the kernel is working just fine, we need
response from venus to handle the statfs syscall, if venus has died or
is otherwise unable to answer requests from the kernel module there
isn't much to do besides complaining loud and annoyingly (fs operations
fail with EIO, messages about failing upcalls in the kernel logs etc.)

Jan
Received on 2003-07-06 16:13:58