Julien Vehent:
> I'm also open to suggestions on how to improve the script. I'm no
> python expert, and there are probably tons of interesting information to
> extract from the postscreen logs.
To improve the script, you could count both the totals and per-client
stats. For my own tiny mail server:
UNIQ/TOTAL EVENT
0/0 BLACKLISTED
0/0 WHITELISTED
0/0 WHITELIST VETO
429/1617 CONNECT
0/0 CONNECT backup
429/1617 CONNECT primary
364/495 DISCONNECT
166/1226 PASS
101/104 PASS NEW
252/376 DNSBL
103/104 PREGREET
2/2 PIPELINING
3/4 NON-SMTP
192/292 HANGUP
7/10 TIME LIMIT
3/3 COMMAND LIMIT
333/493 reject
For example, 429 unique clients made 1617 connections.
This shows that 24% of clients pregreeted, and that only one client
pregreeted twice. The pregreets made up only 6% of all connections.
Below are stats for DNSBL clients only.
UNIQ/TOTAL EVENT
0/0 BLACKLISTED
0/0 WHITELISTED
0/0 WHITELIST VETO
252/384 CONNECT
0/0 CONNECT backup
252/384 CONNECT primary
252/378 DISCONNECT
2/8 PASS
2/2 PASS NEW
252/376 DNSBL
101/102 PREGREET
2/2 PIPELINING
3/4 NON-SMTP
171/269 HANGUP
4/7 TIME LIMIT
3/3 COMMAND LIMIT
0/0 reject
In my case, 40% of blacklisted clients is a pregreeter.
Apparently, the double pregreeter wasn't blacklisted yet.
Wietse
