Just wanted to post some weird, well I find it weird at least, and see
what you guys think?

We made a driver to recursively move some nested OU's about. We did all
the, incredibly tedious, stuff to make every move a single event in the
queue, each event will automatically retry on a warning. This way we can
just wait on the PREVIOUS_MOVE_IN_PROGRESS to complete but now we have
the following:

As soon as the PREVIOUS_MOVE is handled (obituary handled by replica's)
and we move the parent which is empty of all visible objects we receive
an INVALID_REQUEST 641 error. This is the error you receive when you
move a container with leaf objects.

You can actually replicate this without IDM using LDAP just move a three
levels nested container with ldap.

I got this response from support:
"We've done quite extensive troubleshooting on this. The reason why you
get the 641 error is caused by the subordinate objects that were moved
still counting as subordinate objects underneath the OU and flagged as
not present. You can easily see this in iMonitor. This is normal in any
eDirectory tree as moved/deleted objects will create obits that needs to
be processed. Once the obits are processed the old objects still exists
in their original locations but are marked as not present. When the
purger process have succesfully completed after a successful sync, it
will clear all the not present objects with timestamps older than the
last successful sync timestamp. Now once the flags are cleaned and the
not present objects purged, the move goes through without an error.
Thus, the error you're seeing is a timing issue. After moving Test HRSP
21-12-2015, you'll have to wait for the purger process to complete
before you will be able to move Testresultaten. Last but not least, if
you have eDirectory installed on a single server without any other
servers or replica holders, then the not present objects will never be
cleared as the purger process will only run on multiple server TREE's.
This is due to performance reasons."

How can you ever handle this reliably then. If I have to move around
more objects? I don't think you wanna start the purger from dirxml

I would think the PREVIOUS_MOVE_IN_PROGRESS state should be maintained
until purger is done. Also how can you ever move stuff around on a
single-server instance?

Any thoughts, feelings would be appreciated,

Albert-Jan Stevens

ajstevens's Profile: https://forums.netiq.com/member.php?userid=3153
View this thread: https://forums.netiq.com/showthread.php?t=55293