http://defect.opensolaris.org/bz/show_bug.cgi?id=12586
amaguire <alan.maguire at sun.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|ACCEPTED |FIXINPROGRESS
--- Comment #1 from amaguire <alan.maguire at sun.com> 2009-11-10 13:29:34 UTC
---
fix as suggested has been tested and works. "nwamadm show-events" shows the
"old NCP" NCU object state events happening before the NCP state event, which
is then followed by the NCU state events for the newly-active NCP. Note that
the NCP state event is only triggered on init and on an actual NCP change
(simply re-enabling NCP User when it is already enabled won't trigger it). This
prevents reconfiguring everything from scratch.
A few wrinkles
- we were still seeing (online*,conditions unmet) object state events for the
interface NCUs of the old NCP after the switch. The problem is we were trying
to propogate the (online*,uninitialized) state change from the link NCU to the
interface NCU unnecessarily. The fix here is not to propogate state changes to
IP when the aux state is uninitialized - it's unnecessary as we tear down all
interface/link NCUs anyhow.
- we can fill the parent NCP info into the object_state event, but we should do
that when we handle the ncu object state event since (1) it's the only state
event that needs to specify a parent and (2) we have the object lock at that
point so can read the parent from the ncu data directly. To do this, we need to
read in the parent ncp when we populate_common_ncu_props().
Regarding testing, I used a hacked-up copy of nwamadm which shows the parent
NCP for object state events. I verified no object state events for the old NCP
occur after the switch.
--
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
