On Thu, Jun 21, 2001 at 03:53:01PM +0800, Roland Mathis wrote:
> We have a new ghost in our company. It's name is "Loss of Email". We
> have qmail running since about 3 months on our mail server. Every now
> and then people are complaining that they haven't received this and that
> important mail. I usually make a test and it works, but the ghost
did you have a look into your log files when they complained?
> remains. Now, I already installed ISOQLOG, to analyze logs. But I still
> cannot see WHO sent mail TO WHO and WHO receives mail FROM WHOM. Has
> anybody an idea how I could filter this information from the log
> files? Another idea I had is this: Every time qmail-send sends a mail it
Ask for sender, time and recipient first. Ask for the size, too,
and compare this to you databyte limits, if any. Then:
0. convert the time stamp of the log file into something readable,
possibly using tai64nlocal (if logging through multilog is used).
1. grep -i for the sender
Use -i because people rarely get the case right.
If you don't find the sender then grep -i for @domain only -
people _rarely_ get more email addresses right anyway.
You'll find a line like this:
[timestamp] info msg XXXXX: bytes .... from <sen@der> qp ... uid ...
If you don't find it then the mail didn't reach your system, and
you have to look somewhere else, near the sender.
If you find more than one line you'll have to check all entries
or filter for "qp ..." by looking into the senders mail systems
log files (but that is possibly more hassle than it's worth).
For each matching line:
2. Search forward until you find the sender or a line
starting delivery YYYYY: msg XXXXX to (local|remote) [EMAIL PROTECTED]
If recipient doesn't match [EMAIL PROTECTED] (in your eyes) then that's
the wrong message, continue with the next matching "info msg" line.
3. Look what's happened with delivery YYYYY.
I haven't got the slightest idea what ISOQLOG might be, but qmail
log files are simple and there's usually now need for any analysis
tools to find problems (you might want to use them to get statistics).
Robin S. Socha has written a lovely mail, too. Read it and smile.
Regards, Uwe
