On Mon, 1 Dec 2003, Robert Denton wrote: > The post script doesn't have to run in a chroot environment. But it will if > you specify: '%post --chroot' It runs in chroot by default (%post that is). To run in non-chroot specify: %post --nochroot In the non chrooted environment you can find the your drives mounted under /mnt/sysimage. As far %pre, it also is not in a chroot, but I don't think the system mount points are mounted yet either (am I wrong...I will check this in the morning). Anyway if they are not mounted you will need to mount them up where ever you like, copy the keys from there, and then unmount the partitions you mounted. If I am wrong (won't be the first time (-;), you will just need to copy the keys off from /mnt/sysimage. BTW, I think there are a lot of people that want to do this. I know how but if I code it on my company time it won't be opened source (or not anytime soon). Could someone at least start this and GPL what they have. At that point I would be more than willing to help, as this could be used by many others. Also, there a couple of levels of doing this. 1) Save only host keys (A very good start). 2) Save host keys and specified users keys (well really everything in their .ssh directory). 3) Save host keys and all users .ssh stuff. Once you have a mechanism that does any of these (at least 1 of course) the next step is to either: 1) Create an update.img that others can pick and add to their custom anaconda build (A tool to merge multiple such images into one would be nice also). 2) Create a patch to anaconda to allow a special save ssh keys option to a kickstart file. Perhaps, with option one you have made (and like I said the sooner you GPL it the sooner I can help) it very easy to add this to a custom iso image. With this option you would make activities done by %pre be done by a single script and the activities in %post done by another script (or could be the same one called with different args). If this is done as python script or module, the second option can then be easily achieved, which ultimately at least provides the possibility of getting this into the standard anaconda (and I think there are enough users that could benefit from this, that provided the implentation is clean, it would probably be accepted upstream). Anyway, I have to solve this problem eventually, but like I said if I solve it for myself/workplace, it likely won't be made open sourced, unless I can convince my employer that its worth the time spent up front to put it into anaconda directly (in which case I am required to submit the patch back upstream), but knowing our schedules its not likely that things would go down like that. So, since there are so many that are wanting to solve this problem if someone who can do this under the GPL could start something, I think help will come quickly. Cheers...james
