Re: [Exim] Reliability of spool/delivery handling (Linux)?

On Tue, 28 Aug 2001 09:50:19 +0100, Rasputin wrote:

> Ah, no problem.
> I don't think exim does this anyway (although you'd probably
> know for sure, being the maintainer and all).

Well, not on a per-byte basis, but it might be an idea to allow the
administrator to choose to forego the fsync() calls in cases where
machine death is allowed to lose a few messages.

Exim calls fsync()

1) To flush an accepted message body to the spool.

2) To flush the message header to the spool.

2) To flush optional transport output to disk.

3) To flush journal file updates to disk.

Ciao,
Sheldon.