INBOXES\REPLMGR.BOX\INCOMING overfilled

[jhille]

Architecture:

 

Central Site

|

V

Primary Site

| | | | | | |

VVVVV

Secondary Sites

 

 

Background:

 

Collections on the Central and Primary sites were overfilled with a lot of duplicate objects because of VDI provisioning with SCCM agents on our virtual desktops. The number of clients were 50000+. After setting maintenance tasks to clean up inactive clients and aged discovery data 2 weeks old, we were able to clean up our collections. However, this caused a lot of replication problems. Replication was also inconsistent when the collections were overfilled.

 

Problem:

 

The INBOXES\REPLMGR.BOX\INCOMING is overfilled with old faulty replications from the time the collections were overfilled. The number is over 870000 files. This is causing severe hang on the central site every 2 hours. The collections are not updated from the Primary Site to Central site because of this problem. I need to delete a lot of old replication data. How can I achieve this safely?

[Kaido1000]

The main factors that contribute to job backlogs include the following:

 

The distribution of large software packages.

An SMS hierarchy that includes offline sites.

 

See http://support.microsoft.com/kb/836553

 

 

We followed these steps to delete the sending jobs and try to recover the backlog:

 

 

1. Stop the SMS Executive and Site Component Manager services on the site server.

 

2. Run the following command from this folder: <ConfigMgr installation directory>\bin\i386\00000409

 

Preinst.exe /deljob <Site Code>

 

Run this command once for all site codes for sites attached to the site where we have the backlog.

Run it also for the parent site.

 

3. Delete the .srq and .srs files from the following folders:

 

Sms_root_folder\Inboxes\Schedule.box\Requests

-and-

Sms_root_folder\Inboxes\Schedule.box\Outboxes\Lan

 

 

4. Delete the files from "tosend" and "LAN" folders in schedule.box which are more than 6 hours old. As the replication manager would tend to create compressed package files for the jobs which are more than 6 hours old.

 

5. Start the SMS Executive and Site Component Manager services on the site server.

 

6. Now please check site status – component status

Verify there are no errors logged for LAN_sender and Scheduler components.

 

7. Check distmgr.log for any failures on the affected site.