[exim] Exim delivery spiraling out of control

Last night and this morning, I've been struggling with imposing resource
limits on exim's delivery of messages in the spool.

Exim 4.76 (yes, I know it's old and should be upgraded, but that's
another massive problem for me at the moment due to the latest
openssl tickling a compiler bug.)

I have many messages to the same set of hosts in the spool, and each
contains many recipients. (This is a Linux kernel patch set.) What
seems to happen is that exim spawns a great number of processes to
deliver all these messages, even if I use "exim -M <message-id>" to
specify a single message.

I've disabled the queue runner, the exim daemon is currently in
queue-only mode (no -q argument).

I've tried screwing the limits right down, but exim still spawns a great
many processes.

Here's the screwed down resource limits:

remote_max_parallel = 1
remote_sort_domains = *.uk
deliver_queue_load_max = 3
queue_run_max = 1
queue_only_load = 3
queue_only = true

What I noticed last night is that even with this, I get lots of exim
processes with -MC <different-message-id> being spawned.

Is there some way to control exim's resource usage, or is this just the
way exim is?

I've ended up moving the entire exim queue out to a temporary directory,
and moving each message back in, running exim -M <message-id> before
moving on to the next. This seems to be the only way to make exim
behave.

--
Russell King