On Mon, 2011-01-31 at 16:01 +0000, Anne Wilson wrote:
On Monday 31 January 2011 13:18:26 Patrick O'Callaghan wrote:
> On Mon, 2011-01-31 at 12:33 +0000, Anne Wilson wrote:
> > I guess the next step is to take the keychain part out of the script
> > during tests, even though the documentation says that I must have it
> > when running under cron.
> Which documentation? The manpage for crontab doesn't mention keychain
> (it would be wierd if it did). It does say that you can run as a
> non-root user by either a) configuring cron.allow/cron.deny or b) using
> PAM authentication. However the exact conditions under which it uses one
> or the other are not explicit. It may be that Fedora cron always uses
> the PAM stuff, it's not at all clear. It wouldn't be hard to check with
> a toy script though.
The Keychain man page. It specifically says that it must be in the script if
the script is to be run by a cron job. At least, that's what I understand it
I don't even have keychain installed (it isn't installed by default in
Fedora), so I installed it and had a look. I note that the build is from
F12 and the man page is dated 2006. Nothing wrong with that in principle
The point I'm trying to get at is that you don't *need* keychain to get
secure rsync using SSH. It's an optional extra to help manage your SSH
keys. Of course if you're already using it outside this specific case
then naturally you want to keep using it, but if not then there's no
need to jump through hoops to get it to work.