On Mon, Jul 11, 2005 at 05:50:37PM -0500, Chris wrote: > >On a RELENG_5 system from June 30th (source and ports), > >firefox-1.0.4 won't start. It just hangs in the kserel state. > >This is on a diskless system using NFS. > > > >Running truss(1) on the hanging firefox-bin process shows: > > > >kse_release(0x8065fa4) = 383 (0x17f) > >kse_release(0x8065fa4) = 0 (0x0) > >kse_release(0x8065fa4) = 383 (0x17f) > >kse_release(0x8065fa4) = 383 (0x17f) > >kse_release(0x8065fa4) = 0 (0x0) > >kse_release(0x8065fa4) = 383 (0x17f) > >kse_release(0x8065fa4) = 383 (0x17f) > >kse_release(0x8065fa4) = 0 (0x0) > >... > > > >When starting, firefox generates a GNOME Warning: > > > >Gdk-WARNING **: gdk_property_get(): length value has wrapped in > >calculation (did you pass G_MAXLONG?) > > > > C'mon - we two can't be the only ones with this issue...
It's actually quite strange; and seems to be gconfd-2 related somehow. If the file system hosting /var/tmp is a nearly empty memory file system (created by /etc/rc.d/var); everything runs just fine. But if it is NFS mounted; gconfd-2 start before firefox-bin reaches this kselrel state. This gconfd-2 then start polluting /var/tmp with huge amounts of orbit-${USER}-* files. In this case, killing gconfd-2, cleaning up the mess and disabling gconfd-2 like this (I didn't yet find a more sensible way to do it; this is just from memory): $ rm -rf ~/.gconf ~/.gconfd $ rm -rf /var/tmp/orbit-${USER}* $ rm -rf /var/tmp/gconfd-${USER} $ rm -rf /var/tmp/orbit-${USER} $ mkdir /var/tmp/gconfd-${USER} results in firefox-bin starting. It complains that it can't connect to gconfd-2 (which doesn't start), because some file is missing (IIRC something like /var/tmp/gconfd-${USER}/lock/ior), but that doesn't seem to prevent firefox from running without any further problems. I've noticed this by chance, because gconfd-2 kept complaining about permissions of orbit-${USER} files to syslog, generating even more network traffic from the diskless workstation. So, to sumarize: on a clean (memory) /var/tmp file system, gconfd-2 works as it should and firefox starts without problems. On an NFS mounted /var/tmp file system that contains orbit-* files, gconfd-2 starts polluting the directory (and syslog) with huge amounts of orbit-* files (and messages), thus prevening firefox from starting. Preventing gconfd-2 from starting (somehow), lets firefox complaining about it, but it enables firefox to start nonetheless. Very strange indeed. Regards, -cpghost. -- Cordula's Web. http://www.cordula.ws/ _______________________________________________ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to "[EMAIL PROTECTED]"