Problems with setting large replication interval on Site Links

T

Trust No One®

Hi Folks,

I'm after some help. The background is that I'm planning to implement the
SMS 2003 Schema extensions in about 2 weeks time.

We've done a Schema Extension in the past when we upgraded to Windows 2003
and that went very well. This time around I'd like to protect against the
possibility of a Schema extension failure just in case. I know chances of
this are slim but better safe than sorry!

I've done some research and I've decided to go with the method below (credit
must go to Joe Richards who suggested something similar in one of his
many posts):

1) Build a domain controller and place it on a separate subnet from the main
hub site. Create site/subnet objects as appropriate.

2) Create a site link object linking this site to the hub site. Set the
replication interval to a large value say 8640 (6 days) or 10080 (7 days)

3) Temporarily transfer the schema role to this domain controller.

4) Run schema extension program on the domain controller. Once assured that
it has completed correctly, initiate replication with the other domain
controllers outside the site.

I've set this all up ahead of the scheduled schema update in 2 weeks time.
My problem is that the replication interval I've set on the site link object
appears to be ignored when the connection objects are generated by the KCC.

I've set the replication interval to the maximum of 10080, deleted the
existing connection objects to the hub site and chose "Check Replication"
topology to run the KCC. This generated new connection objects. However the
schedule on the connection appears to show replication slots available at 3
hourly intervals (the default)

I've checked with replmon and the domain controller is indeed replicating
with the hub site at 3 hourly intervals (the default for intersite
replication) even though I've specfied a replication interval of 10080
minutes.

So..Has anyone successfully got a large replication interval to work? Is
there anything which I have missed which could be causing this? I could
always create manual connection objects, but I would like to know where I'm
going wrong with the automated KCC.

Best Wishes
 
S

Simon Geary

Rather than going to the trouble of building a DC in a separate site, you
could just disable outbound replication on the schema master before you make
the change. When you are sure the update went smoothly, you can re-enable it
again. This is the tactic Microsoft recommend for the 2003 AD schema changes
and it seems a bit easier than the separate site plan, although I'm sure
that will work as well.

One thing MS definitely recommend against now is yanking the network cable
out of the schema master before applying the schema changes.

To disable: repadmin /options +DISABLE_OUTBOUND_REPL
To re-enable: repadmin /options -DISABLE_OUTBOUND_REPL
 
J

Joe Richards [MVP]

I have found in actual scenarios that you can't push the schedule out more than
one week. I generally try to put that in any posts where I talk about that
solution though I am sure I have probably missed it a couple of times.

There is another solution mentioned by Simon with disabling outbound replication
but that mechanism seems to be a little more troubleprone for folks. I have run
into more than one occasion where someone has indicated that the DC still
replicated. Not sure why it did as it shouldn't, but again, it could be due to
someone making a mistake.

Another possible solution is to put a DC in a site with no site links. Then when
ready, pull it into a site with other DCs.

joe
 
A

Al Mulnick

You know what would interest me most is to know why that don't pull the plug
recommendation is made. I see it in the help files and I see it's been
updated on the KB, but I don't have the background information about why
this recommendation is now changed. Anybody shed some light on that?

Cheers,

Al
 
T

Trust No One®

Simon said:
Rather than going to the trouble of building a DC in a separate site,
you could just disable outbound replication on the schema master
before you make the change. When you are sure the update went
smoothly, you can re-enable it again. This is the tactic Microsoft
recommend for the 2003 AD schema changes and it seems a bit easier
than the separate site plan, although I'm sure that will work as well.

One thing MS definitely recommend against now is yanking the network
cable out of the schema master before applying the schema changes.

To disable: repadmin /options +DISABLE_OUTBOUND_REPL
To re-enable: repadmin /options -DISABLE_OUTBOUND_REPL
Simon,

Thanks for the above. I didn't realise the repadmin command had the above
options. I really must begin work on my Windows 2003 MCSE!

The method certainly seems elegant and I will go with this. Having said this
I'm still puzzled as to why my setting a large intersite replication
interval does not appear to work as in my original post. If anyone can solve
this one I would be grateful as I'd just keep thinking about it otherwise :)

I accept that the probability of a failed Schema update is small, especially
as I've run through the process in our testlab which based on a copy of our
live AD database. Still we are talking of over 100 domain controllers, and
as I'm the one with ultimate sign-off, I wish to ensure a normal length
workday and good nights sleep on the day in question :)
 
D

Dean Wells [MVP]

Probably InitSync ... a requirement that _any_ FSMO must meet before
offering the functions bound to that specific FSMO. The requirement
dictates that, upon boot, the role holder must replicate with a DC
within its own domain (for the domain FSMOs, i.e. - they share the
domain NC) or within its own forest (for the forest-wide FSMOs, i.e. -
they share the config. NC). The DCs must share the afore mentioned NC
as that is the location used to preserve the owner of each of the FSMO
roles (maintained bt the fSMORoleOwner attribute on 5 critical system
objects). This mechanism permits a recently booted FSMO to determine if
the role was seized or forcibly transferred during its absence.

--
Dean Wells [MVP / Directory Services]
MSEtechnology
[[ Please respond to the Newsgroup only regarding posts ]]
R e m o v e t h e m a s k t o s e n d e m a i l
 
G

Guest

I've found that sometimes sites are left in the DEFAULTIPSITELINK and
therefore pick up its replication schedule. Just as wlid guess as I don't
know your setup, but might help.

Gordon
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top