Excessive scratch_buffers_count

the command - watch "/opt/syslog-ng/sbin/syslog-ng-ctl stats | grep ^global"

generates the following line of output. - global;scratch_buffers_count;;a;queued;223338299578

All my other servers have a numeric figure of anywhere between 2 and 500.

I also note that the number of buffer files in the server pushing to this one has masses of files - but only has ONE input - all 01424 buffer files are recorded as no-messages = 0 /syslog-buffer/syslog-ng-01424.rqf', number_of_messages='0'

I've compared configuration files across all our servers, aside from the name of the source, they are pretty much identical, but these two use masses of memory (upwards of 4GB) and have these anomolous status messages.
What can cause these - what am I missing.