Sebastien Roy wrote: > Tony Nguyen wrote: >> Sebastien Roy wrote: >>> Sebastien Roy wrote: >>>> Here's the webrev: >>>> >>>> http://cr.opensolaris.org/~seb/webrev-smf/ >>> I've updated the webrev at the URL above with comments from Alan, >>> Meem, and Liane. I've re-tested, and the updates are copasetic. >>> Please take a second look. >>> >> Looks fine to me. Just a couple nits for net_include.sh >> >> 585 - comments should probably mention net-nwam as another consumer > > Changed to: > > # > # net_reconfigure is called from the network/physical service (by the > # net-physical and net-nwam method scripts) to perform tasks that only > # need to be done during a reconfigure boot. This needs to be > # isolated in a function since network/physical has two instances > # (default and nwam) that have distinct method scripts that each need > # to do these things. > # > >> 609 - use network/datalink-management:default for consistency > > Yes; fixed. > >> 626 - I'd prefer break from if $? -eq 0 and examine $state outside of >> the loop once. > > I'm skeptical about this suggested change. Won't it be possible for the > svcprop command to exit with status 0 after svc.startd has created the > restarter property group but before it has set the state property to > "online"? If so, we still need to wait for the restarter/state property > to be set to "online" by svc.startd by going through the loop again... >
Ah. I believe state is set to uninitialized when restarter pg is created. Yes, there may be a window where state hasn't been properly set. On the other hand, the bigger problem is when state doesn't get set to "online"(e.g maintenance), we'd loop for awhile :) Not too efficient but may not worth the extra logic make it so. I'm fine with the current code :) -tony
