Multiple index rebuilds called if use mongrel cluster?

I use aaf and ferret drb server and one of our indexes is quite large.
In the case that I need to rebuild it this can take up to several
minutes.

The problem I have is that it appears that if you have a mongrel cluster
of 2 or more then if your purge your indexes and restart mongrel that
simultaneous rebuilds can occur which corrupt each other.

I have been forced to only bring up one mongrel, wait for it to rebuild
then bring up the other one.

Have I done something wrong in the configuration or is this expected
behaviour?