Ian Blackwell wrote on Sun, 13 Jul 2008 08:34:51 +0930:
I got similar errors by corrupting my /etc/sysconfig/postgrey file, by putting text into the delay value - i.e. I replaced 660 with 66O. I suggest you recreate the file (from scratch) to make sure you haven't got some odd binary data in their somehow (null's?).
Thanks for the hint. It was the CRLF sequence from creating the file on a Windows machine. I haven't had a problem with this in a long time, bash scripts etc. work fine, no matter if LF or CRLF is used, but it seems to make a difference when including a file.
BTW: Postgrey recommend a maximum delay of 300. Is there a reason you're using 660?
It's the default and been the default since postgrey saw the light of day, but I wouldn't deem it "recommended". ;-) I've been doing greylisting (with sendmail) for many years and started out with ten minutes. This has proven to be quite successful, but there is a growing number of spammers that come back after exactly ten minutes, so I'm moving it up to 11 minutes on new machines. I doubt that 5 minutes gives any advantage in terms of faster turnaround time for ham messages. Most MTAs retry after 15 or 30 minutes, I would actually consider an MTA that retries after only 5 minutes a bit rude.
Kai