Hey Folks, I've been playing around with SMF recently in my spare time. In particular, writing a simple service that will take ZFS snapshots according to a schedule, and do it in a (hopefully) cleanly architected way[1].
There's some blog posts about what I've been doing at: http://blogs.sun.com/roller/page/timf?entry=zfs_automatic_snapshots_prototype_1 http://blogs.sun.com/roller/page/timf?entry=zfs_automatic_snapshots_smf_service and some thoughts on zfs-discuss: http://www.opensolaris.org/jive/thread.jspa?threadID=8643&tstart=60#37156 To summarise, I'm using a transient service that builds a cron job to take automatic snapshots of a given ZFS filesystem (and optionally, it's children), storing the various options for each instance in the SMF repository. I figured that this allows admins to easily manage their snapshot schedules, without having to mess about with cron jobs (yuck). Here's my question: given that the cron job itself is responsible for carrying out the tasks that the service offers, it's disconnected from the usual service logging that the smf-method enjoys. Is there any way to have the cron job log it's output to the service instance log file ? (apart from the obvious re-direction of stdout/stderr) I'd love it if there was some sort of helper-command (or smf_include.sh function) that I could use in a shell script, along the lines of: svclog <FMRI> start . . . svclog <FMRI> stop which would allow me to save all stdout and stderr produced between those commands into the correct SMF log for that instance, without having me having to do all that annoying shell-redirection stuff ... Would this be a valid RFE, or am I missing something obvious that's already provided ? I'd welcome any help (or education via baseball bats :-) that you guys could offer! cheers, tim [1] though that's open to debate! -- Tim Foster, Sun Microsystems Inc, Operating Platforms Group Engineering Operations http://blogs.sun.com/timf
