[OpenAFS] problem with 1.4.5 and Leopard

Derrick Brashear shadow@gmail.com
Wed, 12 Dec 2007 08:37:27 -0500


------=_Part_29870_19184606.1197466647367
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

On Dec 11, 2007 3:27 PM, Sergio Gelato <Sergio.Gelato@astro.su.se> wrote:

> * Sergio Gelato [2007-12-08 18:42:43 +0100]:
> > * Alberto Paoluzzi [2007-12-07 06:48:12 +0100]:
> > > Starting OpenAFS
> > > Loading AFS kernel extensions
> > > kextload: extension /var/db/openafs/etc/afs.kext is already loaded
> > > Starting afsd
> > > /Library/StartupItems/OpenAFS/OpenAFS: line 41:  1158 Bus
> > > error               $AFSD $OPTIONS
> >
> > We've seen this as well, on a G5 that was upgraded directly from 10.3
> > (and OpenAFS 1.4.1). AFS *did* work after the upgrade, then broke.
> > A look at /Library/Receipts reveals that the CHUD package was installed,
> > and since that loads a few kernel extensions (
> com.apple.iokit.CHUDKernLib,
> > com.apple.iokit.CHUDProf, com.apple.iokit.CHUDUtils) it's a rather
> obvious
> > suspect.
>
> We then proceeded to uninstall CHUD, but that alone was not enough to
> get rid of the SIGBUS. Subsequently clearing the AFS cache did the trick,
> however. Unfortunately I didn't think of backing up the old cache
> for analysis, its contents are now lost; sorry about that. Maybe CHUD
> isn't to blame after all. (We haven't tried to reinstall it yet.) I
> suppose
> it's a bug for afsd to crash on a corrupt cache.
>

Probably, yes.
Interestingly it might have been useful to dtruss afsd.

------=_Part_29870_19184606.1197466647367
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

<br><br><div class="gmail_quote">On Dec 11, 2007 3:27 PM, Sergio Gelato &lt;<a href="mailto:Sergio.Gelato@astro.su.se">Sergio.Gelato@astro.su.se</a>&gt; wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
* Sergio Gelato [2007-12-08 18:42:43 +0100]:<br>&gt; * Alberto Paoluzzi [2007-12-07 06:48:12 +0100]:<br><div class="Ih2E3d">&gt; &gt; Starting OpenAFS<br>&gt; &gt; Loading AFS kernel extensions<br>&gt; &gt; kextload: extension /var/db/openafs/etc/afs.kext is already loaded
<br>&gt; &gt; Starting afsd<br></div>&gt; &gt; /Library/StartupItems/OpenAFS/OpenAFS: line 41: &nbsp;1158 Bus<br>&gt; &gt; error &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; $AFSD $OPTIONS<br>&gt;<br>&gt; We&#39;ve seen this as well, on a G5 that was upgraded directly from 
10.3<br>&gt; (and OpenAFS 1.4.1). AFS *did* work after the upgrade, then broke.<br>&gt; A look at /Library/Receipts reveals that the CHUD package was installed,<br>&gt; and since that loads a few kernel extensions (com.apple.iokit.CHUDKernLib
,<br>&gt; com.apple.iokit.CHUDProf, com.apple.iokit.CHUDUtils) it&#39;s a rather obvious<br>&gt; suspect.<br><br>We then proceeded to uninstall CHUD, but that alone was not enough to<br>get rid of the SIGBUS. Subsequently clearing the AFS cache did the trick,
<br>however. Unfortunately I didn&#39;t think of backing up the old cache<br>for analysis, its contents are now lost; sorry about that. Maybe CHUD<br>isn&#39;t to blame after all. (We haven&#39;t tried to reinstall it yet.) I suppose
<br>it&#39;s a bug for afsd to crash on a corrupt cache.<br><div></div></blockquote><div><br>Probably, yes.<br>Interestingly it might have been useful to dtruss afsd.<br><br></div></div><br>

------=_Part_29870_19184606.1197466647367--