>Great-- I saw a significant problem when I was at JavaOne with
>a set of (50) machines that had a troublesome custom USB device
>driver-- so, you'd plug in the USB device, the driver would
>wedge or crash a little later, and then you'd have to reboot *and* cope
>with the now dirty boot archive.
Indeed, I had a headless system and laptops wedge this way
on various occasions.
Headless option:
svccfg -s boot-archive setprop start/exec = ":true"
svcadm refresh boot-archive
and Bob's your uncle.
>Overall, I have found the boot-archive-is-out-of-date handling to not be
>the most friendly of user experiences. Anything we could do to make it
>a little more guided would be helpful I think. The "ticking time bomb"
>aspect is not my favorite, either-- if an innocent sysadmin plugs a
>USB device into a server to load up some files, unplugs it, then
>6 months later the machine crashes, then the boot archive is going
>to be out of date and the machine is not going to cleanly and
>automatically restart itself-- which could increase the customer's
>downtime.
>
>Once a day cronjob to update the boot archive?
Ignore failure after panic?
My favourite issue is the timezone changeover; (rtc)
Then again, I run all my systems with the clock in localtime.
Casper
