Optimization during large migration?

In running my vss2svn script, I do notice sometimes that it slows down
quite a bit after it runs for a long time. It's during the actual "svn
commit"s that the slowdown occurs; i.e. committing a few small changes
takes longer and longer as time goes on.

My first instinct was to dramatically increase the in-memory buffer and
log file size to something like 16MB and 64MB, but it seems as though a
memory buffer any bigger than 2MB causes "svnadmin recover" to fail with
"svn: Invalid argument" (although I couldn't find anything about a 2MB
limit in the BerkleyDB docs).

Can anyone suggest some general strategies for such a specialized case,
where the server is bombarded by one commit after the other for hours on
end? Is there any sort of optimization I can do on Apache itself?

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Fri Jun 25 00:00:59 2004