On Tue, 30 Sep 2014 20:56:35 +0200 Jaap Winius <jwin...@umrk.nl> wrote:
> In the mean time, I still get the most consistent results by using two > separate start-stop-daemon invocations. I understand that this is not > ideal, since now the script will start the long-term daemon even if > k5start fails before it, but right now I don't have a better solution. I haven't read through this thread too closely, but it doesn't seem like you've been given much of a concrete explanation of how you're supposed to change a daemon startup invocation to a k5start-ified one. (But there has been a lot of attempted explanation of what's going on or why something doesn't work.) So I'm not going to answer your immediate question, but just to try to maybe help illustrate: If you have an init script that at some point runs some long-running command in the foreground (after forking etc), like: /usr/sbin/whateverd --foreground You should replace that with: k5start -f /whatever.keytab -U --other-options -- /usr/bin/whateverd --foreground (or /usr/bin/k5start, or $K5START or something) But if the init script runs some command that daemonizes itself (it forks, detaches, etc), you need to replace this: /usr/sbin/whateverd --daemonize with something like: k5start -b -f /whatever.keytab -U --other-options -- /usr/sbin/whateverd --foreground Note the '-b' option for k5start, and changing the whateverd invocation from '--daemonize' to '--foreground'. If the long-running process does not have a way of running in the foreground, I don't think you can do this the 'proper' way that people have been suggesting. And of course, all of this is just the arguably 'proper' way to use k5start; that is, to wrap another command. You can just run k5start separately and be decoupled from the other process, as I think you are doing, and other people do do that. It's just not the "best" way to do it (at least, according to some/most). Also, if you want more feedback or more eyes on this or anything, you can try mailing the kerberos list, since this isn't really AFS-specific. I don't mean to say that this thread is inappropriate for here, but just that that's another place to try if you wanted. Others may be more used to answering questions like this there, and Russ would see it there (iirc he doesn't read this list, or will not be reading it for long). -- Andrew Deason adea...@sinenomine.net _______________________________________________ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info