VHCS ->iscCP mess..

[Zeitkind]

Hi,
after checking the progress of ispCP for some months, I finally decided to upgrade.
One reason was a Debian sarge (601 days uptime *sigh* which needed to get updated anyway and the lack of support for VHCS2.

What I had:
A working VHCS2 with some little mods:
VHCS2 did nothing to my bind files, I have ~25 domains running, but prefered to make the host-files & config (eg. slave) by myself. It was simply a "do nothing"-hack to the script. No users added domains by themself, it is a more private server.
postfix config was completly rewritten to add more RBLs, SPF and such.
I did nothing to the "main" VHCS2 things like db-layout etc.

What I did:
- Upgraded sarge AMD64 to etch without much problems. All went fine.
- checked the packages ispCP needed - all fine
- run setup & upgrade script - all fine

Looked quite good. But...
- all sites had a ispcp-login instead of the content
- DNS totally screwd - expected that, I could have fixed this
- no POP or IMAP login was possible at all
- Amavis, spamassassin & policyd screwed and completely broken
The only thing that was running was postfix

I googled around, did not find much help, so I decided to do a "half-clean" install. I purged bind, apache, courier, postfix etc. - all I kept was mySQL. Did a mySQL dump of ispCP-tables, trashed the /etc/<package> dirs and reinstalled the services. Started over with iscCP setup and found an empty db. Imported the ispcp.sql and set all status to toadd. My users and domains came back.. fine.
Noticed that https, pop-ssl, imap-ssl, policy-weight and spamassassin were not running. Seems the debian-packages-etch miss some stuff like courier-ssl, installed them and services were up. Apache-ssl configured simular to the hint here somehwere - fine. Now all services are running, sites show their content again, postfix is running - but..
Still can't login to pop3 or imap.

May 22 00:39:40 server couriertcpd: authdaemon: s_connect() failed: No such file or directory
May 22 00:39:40 server couriertcpd: authentication error: No such file or directory
in mail.err
Sending emails seems to work:
May 22 04:31:44 server postfix/smtpd[13168]: connect from p4FD8F0A9.dip.t-dialin.net[79.216.240.169]
May 22 04:31:45 server postfix/smtpd[13168]: 31103EAC120: client=p4FD8F0A9.dip.t-dialin.net[79.216.240.169], sasl_method=NTLM, sasl_username=info@xxxxxxxxx.de
May 22 04:31:45 server postfix/cleanup[13203]: 31103EAC120: message-id=<8FC251CE-4155-4D32-AD2A-E767742EBA29@xxxxxxxxx.de>
May 22 04:31:45 server postfix/qmgr[9465]: 31103EAC120: from=<info@xxxxxx.de>, size=599, nrcpt=1 (queue active)
May 22 04:31:45 server postfix/virtual[13213]: 31103EAC120: to=<xxxxxxxx@xxxxxxxx.com>, orig_to=<postmaster@xxxxxxxx.com>, relay=virtual, delay=0.22, delays=0.21/0/0/0.01, dsn=2.0.0, status=sent (delivered to maildir)
May 22 04:31:45 server postfix/qmgr[9465]: 31103EAC120: removed

Spamassassin doesn't work too, but I guess it's simply not configured like clamav.
I never dealed with courier before, so I have absolutely no idea where to look at, any hints/suggestions are welcome.

[kilburn]

Try editing the "/etc/courier/authdaemonrc" file and settting "DEBUG_LOGIN" to "1" or "2". Then restart courier and try to connect, the logs should show more info than that so we can further diagnose...

[Zeitkind]

kilburn Wrote:Try editing the "/etc/courier/authdaemonrc" file and settting "DEBUG_LOGIN" to "1" or "2". Then restart courier and try to connect, the logs should show more info than that so we can further diagnose...

Already tried:
server:/etc/init.d# ./courier-authdaemon reload
Stopping Courier authentication services: authdaemond/etc/courier/authdaemonrc: line 68: -: command not found

line 68 was the uncommented debug-level command.. well.... ermm.. huh?

[Zeitkind]

Upps.. I forgot to clear the rest of the line...

Found the problem, only pam was active, not userdb for authdaemonrc

Edit:
Now we were able to activate amavis & clamav, included dul.dnsbl.sorbs.net into policy, added some more options to postfix again like
check_client_access cidr:/etc/postfix/postfix-dnswl-permit, (a whitelist with all major ISP's MTA's)
heck_helo_access hash:/etc/postfix/ok-helos, (manual edited list for broken MTAs like "Exchange.local")

and fixed some other stuff. Changed bind-script for spf from ~a to -a (I prefer fail not softfail). Things left to check are ftp and such.