Hi, I'm sure this is a newby problem but hopefully someone can help.
We have a SQL 2000 merge replication scenario here with one publisher and 25 subscribers. Recently I needed to remove an article from publication, so I dropped all subscribers to drop the article. When adding all subscribers again, the initial synch was painfully large and slow. It prompted me to investigate why. I ran the following query: SELECT COUNT(TableNick) FROM msMerge_Contents and the result was over 25000.
I have confirmed that the Agent Profiles for all subscribers have MetadataRetentionCleanup = 1. My understanding is that when this value is set to 1, it automatically cleans up the merge system tables.
Also, while researching for an answer I also discovered that nearly all our subscribers tables failed row validation (this was manually done and not scheduled by the agent). Through Enterprise manager, I tried 'Validate and synchronize', but that doesn't work. It's not a good feeling knowing that not all data is equal amongst the computers.
Are we doing something wrong? How can we reduce the merge system table sizes to something reasonable and maintain that? How do we fix the invalid (or missing) data?