rid master during admt

[Graham Turner]

just wanted to get any issues that have come to light during an ADMT process
specfically relating to the function of the RID master

when doing a mid size migration (approx 4000 objects) it seems the
allocation of RIDs' may become an issue.

the operation of rid master is understood to be allocation of a new pool
after a percentage of the current pool is exhausted -

in an ADMT 2.0 migration of approx (4000 users) this would be quickly
exhausted perhaps leaving potential for failure of the ADMT 2.0 if it can
get another RID in good time from a newly allocated pool.

this does beg another question - is there any known issue of delay in the
allocation of new rid pool ??

GT

 

[Eric Fleischman [MSFT]]

Hi Graham,

You raise an excellent point. For every new object we create that we assign
a SID to we do need a RID.

My recomendation would be to ensure that you have the RID master online and
accessable during the allocation. You can check your current RID allocation
if you'd like by running a dcdiag /v on the DC in question. That shows the
current RID pool, but it clearly won't be large enough for 4000 users.

So, you want to ensure the RID master is online and operating.
In terms of latency, I woulnd't worry. A DC can get a fresh set of RID's no
problem and it is a very fast operation. It should not cause issues with
your migration. I've migrated 25K users (or was it 50K? it was a lot, I
forget) in one shot in production and never have had a RID issue so long as
my RID master was online.

Oh, and of course if you're migrating on DC's in the US having it on a dc
that is "close" is better, but certainly not required. Having it near the
DC's where you'll be doing the migration is adventageous, but again not
required by any means.

Hope this helps!
~Eric