readonly-root with a non-readonly-root

Jeffrey Law law at redhat.com
Thu Aug 31 08:10:44 UTC 2006


On Wed, 2006-08-30 at 23:08 -0500, Paul B Schroeder wrote:
> /etc/rc.d/rc.sysinit currently gets the relies on getting $READONLY from 
> /etc/sysconfig/readonly-root.  It uses this to determine whether or not to call 
> /etc/rc.readonly and whether or not to remount the root filesystem in read-write 
> mode.
> 
> In some instances (think flash based filesystem), it would be desireable to run 
> the rc.readonly script (we don't want to burn up the flash), but still have the 
> root filesystem in read-write mode (we still want to be able to do some 
> filesystem editing).  Maybe rc.sysinit could check two separate variables to 
> control this?
> 
> $EXEC_RC_READONLY or whatever you would want to call it, could be used by 
> rc.sysinit to determine whether or not to exec rc.readonly.  And $READONLY could 
> still be used to determine whether or not to remount root read-write.
> 
> Thoughts?
As Mark mentioned.  All this has changed.  The capabilities of rc.readonly
have moved into rc.sysinit.  It will probabl continue to morph as we
continue working on stateless linux.

We also have some capabilities (via bind mounts) to allow certain
files/directories to be read/write while the root filesystem itself
continues to be readonly.  You could probably use those capabilities
to accomplish your task, particularly if you can enumerate every
file/directory which you want to be writable.

Jeff




More information about the fedora-devel-list mailing list