"add all tables" slows over time on many tables

[michaelcrobinson]

I guess this is less of an issue, more of an observation, since my use case is a little weird.

I have a database with over a hundred schemas across over a hundred tables. Needless to say, this results in a lot of tables, and I knew it would take a long time to run "add all tables". Something I noticed is that over time, it slowly started taking longer to add tables. Two hours in, the process had slowed to 1 tables every >1s. It also appears to be single threaded, not that that matters.

As a workaround, I wrote a script that queries my database for a list of schemas, and for loops though the list to do all tables like so:

for SCHEMA in $SCHEMAS ; do
  bucardo add all tables db=source_db relgroup=db_group -n $SCHEMA
done

Works for add all sequences too.

[turnstep]

I was able to duplicate this problem: the slowdown occurs in the repeated INSERTs to the herdmap table. Specifically, the herdcheck trigger was slowing things down too much. Made a proposed fix at #234