Ken McDonell wrote:
On Tue, 2009-09-29 at 08:39 -0500, Martin Hicks wrote:
...
How much of the cluster PMDA must be done now?
I suspect that we *must* at least update the duplicated data structures
with the larger domain bit width. I think the optional part would be to
also start using the new dynamic namespace stuff...
Unless the cluster PMDA is assigned a domain number > 255 (seems
unlikely at this stage), it "should just work" (tm) ... the low-order 30
bits of a pmID and pmInDom are unchanged with the PMDA_INTERFACE_4 round
of changes.
OK, that's easy enough to test, when I get a tick. I think I can
smell a new QA test cooking ;-)
Now you may _want_ to make changes to the cluster PMDA but I'm not sure
they are necessary before declaring 3.0 ready to go ...
ok, we can do that, depending on the results of the test
indeed one could
argue that 3.0 should happen first then the PMDA_INTERFACE_4 changes are
merged ready for a 3.1 release.
I'm still running QA, it's awfully slow in a VM. For the sake of relieving
the constipation, I'll push all of this to the dev branch tomorrow - my
stuff first (3.0.0), then with Ken's tree merged in too (3.0.1).
Cheers
-- Mark
|