Replication schedule vs. interval ????

[Sandy wood]

I'm getting alerts in my MOM 2001 that my sites are
taking longer than 15 minutes to complete intersite
replication - sometimes 50-111 minutes. I've got the
interval set to 15 minutes and the schedule at 4 times an
hour. My site connections are ATM speed and no more than
8 miles from our central site. All outer sites replicate
to and from a central site. I can force a replication and
then see in replmon that it suceeded within 2-3 minutes.
Where is the holdup that MOM might be seeing?

 

[Joe Richards [MVP]]

Could be a zillion things.

When it starts reporting that you should check the queues on the bridgehead servers. You can do that with perfmon
counters specifically NTDS\DRA Pending Replication Synchronizations which you have to watch out for as there are some
bugs in some of the post-sp3 and pre-sp4 versions. If it doesn't ever come back to zero it could be the bug. I don't
know if ever got the KB Article number that was assigned to that bug when I found it and forwarded it to MS.

Alternatively you can use repadmin /queue to look at each queue to see if they are backed up and what specific items are
in the queue. Also I have a tool on my website - www.joeware.net called adqueueloop that will do autorefresh display of
a queue and how many items are in it and what item is on the top. If you have a replication item that sits there for
more than 30-40 seconds, it probably indicates a DC that is having issues whether it is network or machine related takes
further troubleshooting.

 

[Sandy Wood]

Thanks for the help in drilling down to some specifics on
this one. I haven't seen the MOM alerts since I dialed
down my replication interval / schedule to 15 minutes for
both. I'd still like to see what's in the queues and for
how long. I've used repladmin to force a replication
between sites and it shows being successfull in a minute
or so, so I'm probably ok on that, but I'll check out
that tool on your site.



Thanks again for the help.

-----Original Message-----
Could be a zillion things.

When it starts reporting that you should check the

queues on the bridgehead servers. You can do that with
perfmon

counters specifically NTDS\DRA Pending Replication

Synchronizations which you have to watch out for as there
are some

bugs in some of the post-sp3 and pre-sp4 versions. If it

doesn't ever come back to zero it could be the bug. I
don't

know if ever got the KB Article number that was assigned

to that bug when I found it and forwarded it to MS.

Alternatively you can use repadmin /queue to look at

each queue to see if they are backed up and what specific
items are

in the queue. Also I have a tool on my website -

www.joeware.net called adqueueloop that will do
autorefresh display of

a queue and how many items are in it and what item is on

the top. If you have a replication item that sits there
for

more than 30-40 seconds, it probably indicates a DC that

is having issues whether it is network or machine related
takes

further troubleshooting.

in message news:[email protected]...

 

[Joe Richards [MVP]]

No problem.

--
Joe Richards
www.joeware.net

--

Thanks for the help in drilling down to some specifics on
this one. I haven't seen the MOM alerts since I dialed
down my replication interval / schedule to 15 minutes for
both. I'd still like to see what's in the queues and for
how long. I've used repladmin to force a replication
between sites and it shows being successfull in a minute
or so, so I'm probably ok on that, but I'll check out
that tool on your site.



Thanks again for the help.
queues on the bridgehead servers. You can do that with
perfmon
Synchronizations which you have to watch out for as there
are some
doesn't ever come back to zero it could be the bug. I
don't
to that bug when I found it and forwarded it to MS.
each queue to see if they are backed up and what specific
items are
www.joeware.net called adqueueloop that will do
autorefresh display of
the top. If you have a replication item that sits there
for
is having issues whether it is network or machine related
takes
in message news:[email protected]...