Darrick Hartman (lists) wrote: > Philip Prindeville wrote: > >> Fred wrote: >> >>> Lonnie >>> >>> At 12:31 26/01/2008 -0600, Lonnie Abelbeck wrote: >>> >>> >>>> Well, the unionfs "rw" overlay to the "ro" base mount is CF based, but >>>> most any files that are written-to on a constant basis are mounted as >>>> "tmpfs" which is a RAM based filesystem. >>>> >>>> >>> OK. So am I right in thinking that for development, I save files to /tmp >>> like there's no tomorrow since those write's are actually made in RAM, and >>> only save the program to eg. /usr/local/bin once it's debugged? >>> >>> In other words, if I want stuff written to /var to survive a reboot, I must >>> edit /etc/fstab to remove the line for /var, copy files from /var to >>> /dev/hda2 somehow, reboot, and from then on, AstLinux will use hda2 to hold >>> /var? >>> >>> Here's what I have after adding /dev/hda2 as UnionFS: >>> >>> ====== >>> pbx ~ # mount >>> rootfs on / type rootfs (rw) >>> /dev/root on /oldroot type ext2 (rw) >>> /dev/hda1 on /oldroot/cdrom type vfat >>> (ro,fmask=0022,dmask=0022,codepage=cp437,iocharset=iso8859-1) >>> >>> >> ^^^^^^^^^^^ I'm getting this too, and it's really annoying. >> >> I have to modify my script that pushes a new release of the OS to first >> unmount this, otherwise when I go to remount /dev/hda1 it tells me that >> it's busy. >> >> Is there a simple fix for this? >> > > Philip, > > This is not related to what Fred is asking. Fred wasn't talking about > /oldroot/cdrom remaining mounted. He was talking about var (which does > not reside on unionfs). >
I know he wasn't... but since it came up in the example, I thought I would ask about it. Wasn't sure if it was deliberate or if it was a hald/udev oddity... Modified the installer-script. I'll post it if anyone is interested once I get the bugs out. BTW: What's necessary to install onto this filesystem multiple instances of the OS so you can have a fallback version of the OS? Every time I install, I first zap the old version of os/ ... since there's room for two or more images on a 128MB partition, we might as well keep a fallback version. > If you don't have enough system memory, you must leave /oldroot/cdrom > mounted. That's because the image file is still mounted loop back. If > you have enough system ram that the image file is copied to a ramdisk, > it should be possible to unmount /oldroot/cdrom. I say should because > whenever I've tried, the system says there's a file in use and it can't > be unmounted. > How do you know if you've had enough memory or not? What are the external signs? I don't have a serial console connected all the time, so I lose whatever boot-time messages their might have been. If it does make a ramdisk copy, can it deliberately unmount the partition so we know it doesn't need it any more? -Philip > Since I have not seen your updating script, I can only make the > following suggestions for now. Either assume it's already mounted and > just issue 'mount -o,remount rw /oldroot/cdrom' or check to see if it's > mounted, then run the correct command depending if it's mounted or not > mounted. > > As to Fred's persistent /var question, I would not want /var using the > cf for everything. You could add some symbolic links (add your script > to rc.local on /mnt/kd) at startup to link your partition from > /var/<dir> to /mnt/kd/<dir> > > Darrick > ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Astlinux-users mailing list Astlinux-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/astlinux-users Donations to support AstLinux are graciously accepted via PayPal to [EMAIL PROTECTED]