Hi Rune,
Thanks for your help with this.
On Sunday 02 March 2014 07:28:03 u-codalist-r...@aetey.se wrote:
> Hello Andrew,
>
> On Sun, Mar 02, 2014 at 01:01:48AM +0000, Andrew Wasielewski wrote:
> > I am doing a fresh install of Coda client on Fedora 20. When I run
> > coda-client-setup (Fedora name for venus-setup) I get the following error:
> >
> > [root@ivanka-laptop ~]# coda-client-setup server.wasielewski 20000
> > /usr/sbin/coda-client-setup: line 176: [: char-major-67-*: binary operator
> > expected
> >
> > That part of the script adds aliases for the coda kernel module to
> > /etc/conf.modules. The problem seems to be the "*" in the modinfo output
> > on FC20:
> >
> > [root@ivanka-laptop etc]# /sbin/modinfo -F alias coda
> > char-major-67-*
> > fs-coda
>
> The problem arises because modinfo outputs multiple values there.
> This is a bug in the upstream venus-setup script: missing quotation
> marks in
> [ -z $(/sbin/modinfo -F alias coda) ]
> which should be
> [ -z "$(/sbin/modinfo -F alias coda)" ]
> (the problem is hidden as long as there is no more than a single alias)
>
I noticed after I posted this that even with this problem, the rest of
coda-client-setup completes successfully. With the '"' fix the error message
goes away, though it doesn't build /etc/conf.modules. However even if it did
it wouldn't achieve its purpose, as Fedora at least doesn't load modules that
way. You have to create a .module file in /etc/sysconfig/modules (see
http://docs.fedoraproject.org/en-US/Fedora/18/html/System_Administrators_Guide/sec-Persistent_Module_Loading.html).
This seems to work well for me:
[root@ivanka-laptop modules]# cat coda.modules
#!/bin/sh
if [ ! -c /dev/cfs[0-4] ] ; then
exec /sbin/modprobe coda >/dev/null 2>&1
fi
It looks like someone raised this as a bug on Fedora a while back:
https://bugzilla.redhat.com/show_bug.cgi?id=450463.
> > Should I raise a bug report about this? I guess I could manually create
> > the conf.modules from modinfo output and comment out this section of the
> > script.
>
> I have now created a ticket for upstream Coda. Nevertheless, it is
> certainly good if you report it to Fedora, the Fedora maintainers may
> wish to apply this change to their package.
>
I also raised a couple of other bugs with the Fedora packaging of server-side
components:
https://bugzilla.redhat.com/show_bug.cgi?id=1071534
https://bugzilla.redhat.com/show_bug.cgi?id=1074321
These both relate to the Fedora systemd integration, not the underlying coda
functionality.
> Otherwise I wonder whether the Aetey's universal installer would work
> for you (it does not try to configure kernel module loading, beyond an
> attempt to modload coda at the Coda client startup)
> http://www.aetey.se/dl/cocli-2.1-6.9.5-linux-ia32-3.bin
> md5: b2897b0be53b1b209f85fdf0a5452504
> http://www.aetey.se/dl/cocli-2.0-6.9.5-linux-x86_64-0.bin
> md5: 971080aeece1829b4730021337761342
> It does not obey any startup mechanics automatically, you have
> to let the distro's tools run "codaclient start" at boot time
> ("codaclient stop" for a clean shutdown, not critical) or do this
> manually as root.
>
> Regards,
> Rune
>
Thanks & regards,
Andrew
