[quote:03a7376c0a]
1. Bandwidth for HTTP domains works
2. Bandwidth for subdomains is not recorded
[/quote:03a7376c0a]
I have a problem with this. I just installed mod_gzip and this is what I found:
- subdomain and domain access_logs seem to record actual data transfer, except for HTTP headers.
- bytes_log logs seem to include headers size, but before compression.
Headers could add up to a lot at the end of the month (even if they are only the response headers), so relying on the access logs alone is not a very good idea. And neither would be relying on the bytes_logs, because the clients will see the stats of webalizer or analog (which take their data from the access_logs).
I don\'t think I want to disable mod_gzip. Could the bytes_log module be fixed/tweaked to address this problem? Does the BytesLog directive accept any parameters?
[quote:03a7376c0a]
3. Bandwidth for FTP is broken
[/quote:03a7376c0a]
What about this:
LogFormat FtpCount \"%v %u %b\"
<Global>
ExtendedLog /var/log/proftpd.log ALL FtpCount
</Global>
With these two directives server wide we could control the ftp server (%v) or the name based domain users (%u) and the bytes transferred (%b) for all FTP requests. Would this impose too much unnecessary overhead on the server?
[quote:03a7376c0a]
4. Bandwidth for SMTP is not recorded (might be possible)
[/quote:03a7376c0a]
Exim logs include message size, for both received and sent postings (the S values at exim_mainlog). Could that be used, or are there better ways to measure this?
[quote:03a7376c0a]
5. Bandwidth for pop is not recorded (bad.. we should be able to do this in theory)
[quote:03a7376c0a]
I\'ve also added support for pop bandwidth tracking...
[/quote:03a7376c0a]
[/quote:03a7376c0a]
Are these data available somewhere on the server (a log file that I can read)?
[quote:03a7376c0a]
6. Bandwidth for all other protocols is not recorded ?? what other protocol are there though (that do real bandwidth) ??
[/quote:03a7376c0a]
I would like to know that too. The acctboth chain also counts bytes for ports 2082/2086 (cpanel/whm), 6667 (ircd) and 3306 (mysqld).
By the way, in my server it doesn\'t count the bytes for each port, only globals (any->any) for \'tcp\', \'icmp\', \'udp\' and \'all\'. It used to, but no longer does... I don\'t know why

and I have no idea of how I could fix it - any hints?
And it would be great if mailman data transfer were assigned to the domain owning the list. It would also be great if mailman could be translated to other languages, because it\'s all in English and people here tend to use their own mailing list scripts, sending all mail as nobody. That\'s bad, and I can\'t translate mailman files because upcp will overwrite them in the morning

Suggestions welcome...
Juanra
[Edited on 12/27/01 by Juanra]