On 03/24/10 03:46 PM, Joseph J VLcek wrote: > On 03/23/10 06:26 AM, Joseph J. VLcek wrote: >> Hey Jack, (or anyone else who would know) >> >> I am adding a new DC finalizer script for populating a pkg(5) package >> with the AI image bits. >> >> I had hoped to take advantage of the temp directory to pass a file of >> data from an earlier checkpoint to the new checkpoint. >> >> However I have noticed that between checkpoints the temp directory, >> passed to every finalizer script, is cleaned up. >> >> Are the temp directories cleaned up between checkpoints by design? >> >> Thank you. for any and all help. >> >> Joe >> >> >> _______________________________________________ >> caiman-discuss mailing list >> caiman-discuss at opensolaris.org >> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss > > > Just to follow up. > > I had spoken with Jack about this on the phone. > > Currently when the distro constructor finishes it removes the contents > of the tmp directory. This is true even it the dc is paused at a > specific checkpoint. When resumed the contents of the temp directory are > removed. > > The one line of code which removed the contents of the temp directory is > in the main_func() of distro_consts just before return: > > > dcu.cleanup_dir(cp.get_build_area_mntpt() + TMP) > return (status) > > The proposed changed is to remove the dcu.cleanup_dir call and add a new > finalizer script which can be run as the very last finalizer script > which will clean up the contents of the temp directory. > > If anyone has a different suggestion please pass it along. >
I'd suggest instead that "distro_const build" when invoked without -R or -r (in other words, starting from the beginning) should clean it up, rather than doing it on exit or relying on it being a specific finalizer script. Dave
