Hope that anybody can tell me something about this problem. I have a active Replication (RTB) in continous mode with First + Partner site. If the system perform a cleanup operation, CPU utilization increase to 100%, 50% to ntdbsmgr, 50% to replserv. There are a lot of records to purge/clean (n.b. 7000000). Cleaning interval is default (7days). Is there any possibility to increase this process or prevent the system from the use of 100% CPU utilization ? (System is active 24/7, so it is difficult to handle changes)

 Many thanks for all informations.

 Are you really going through 7M records in the database a week?  Or have you had to rebuild the file and run DXSYNCTABLES on it to re-sync everything?

Our DXUtil tool (www.goldstarsoftware.com/tools.asp) allows for a manual purging of the PDC obituaries (DXUTIL -PURGE) directly from the Btrieve level, which is typically faster and contains much less overhead.  At this time, though, it deletes ALL obituary records, meaning that it should ONLY be used when you know that the records are in sync between the two sites. 

Of course, users deleting records as the purge is running may have their obituaries cleaned out, too, resulting in some data NOT being synchronized correctly, so this should be done during a period of no activity (if possible).  If not, then DXSYNCTABLES afterwards can clean it up after the fact.

If you need a real "cleanup replacement", then let me know and we can investigate the possibility of adding a datestamp cutoff as well.
 

Have you considered decreasing the values of when the cleanup is done and how long the deleted records are kept.  This can be set using the dxcleanup utility.

Jose Pruneda

The problem with 7M records occur, if we have setup a new parter site, and the default cleanup is working in a cycle of 5 days, so near 1 week after the initial setup the cleanup of a big blog runs.  The normal changes/adds in this database are 1M records in the database a week. We have another database in sync with more than 2M changes/add per week, but in this case the sync. works on each hour and servers are more powerful, so we don't run in this problem. Here in this case, the sync is working continously and computers are not so powerful, so if the cleanup runs and allways the system try to run the sheduled sync and users are active, the performance is not high enough to service all parts.

 So i will check if a "cleanup replacement" is a solution in this case.

 Many thanks.