How to recover from a messy situation?

[aglien]

Hello all,
I'm in a trouble.
I was using ispcp 1.0.0RC6 but, step by step, it started to working bad and I started to change the /etc/apache2/sites-avaliable/ispcp.conf by hands.
Finally I tried to upgrade to version 1.0.0.
I followed this tutorial:
http://www.isp-control.net/documentation...upgrade:rc
All was fine until I launched this script:
perl ispcp-update
After a while it stopped working with the error: "mysql was gone away".
I changed the mysql server time-out and I relaunched the script.
This time the script stopped simple with "failed".
Now all is very messy. Ispcp is not working.
The server is a production server. It contains about a hundred sites. The sites are working but now is impossible to change anything.

Is it possible to recover all (or somethings) by hands?
Is it possible to transfer the sites, one by one, to another server?
Any hint is welcome.
Many thanks

Enrico

[kilburn]

Indeed, you're in a tough situation :S

Quote:Is it possible to recover all (or somethings) by hands?

Do you have any backups? You may try to reinstall rc6 and restore the db backup so you get a working panel again. To prepare a "full recovery plan" I would need more information (what kind of changes have you made to apache's config? Can you loose them and manually re-apply them later? What kind of errors are you getting now?).

Whatever the case... backup first!

Quote:Is it possible to transfer the sites, one by one, to another server?

AFAIK, there's no *easy* way to do it. You can work out a migrator based on the cli managing script made by aseques though.

Hope this helps...

[aglien]

(01-21-2010 05:42 PM)kilburn Wrote:  Indeed, you're in a tough situation :S

:'-(

Quote:Do you have any backups?

I have backups of:
/usr/sbin/maillogconvert.pl
/var/log/*
/var/mail/*
/var/www/*
/etc/*

but I forgot to do a db dump :'-(
and the problem is that the situation was bad also before my attempt.

Quote:You may try to reinstall rc6 and restore the db backup so you get a working panel again.

Is there somewhere any automatic db backup?

Quote:To prepare a "full recovery plan" I would need more information (what kind of changes have you made to apache's config? Can you loose them and manually re-apply them later?

It seems that the biggest problem was (and is) the file:
/etc/apache2/sites-available/ispcp.conf
Ispcp messing up it many times, duplicating items and so on.
From the panel was impossible to change the status of a domain.
The same was with /etc/bind/named.conf but this is simpler to maintain by hands.
Now I fear to start again the ispcp daemon.

Quote:What kind of errors are you getting now?).

From the point of view of the panel, quite all domain are in a bogus status.
For example, after this command:
select domain_name, domain_status from domain;
I get this:
get_tag() | ERROR: '# httpd [{IP}] virtual host entry BEGIN.
' ne '# httpd [{IP}] virtual host entry END.
', '# httpd [{IP}] virtual host entry BEGIN.
' or '# httpd [{IP}] virtual host entry END.
' missing in src !

Anyway, ispcp knows its status:
"We are sorry, but the system is currently under maintenance.
Please try again later."

Quote:Hope this helps...

Many thanks

---

One important thing to add: the fqdn hostname was changed!
This could be one of the reason of my problems.
What I have to change in the configuration files to change the hostname?

[kilburn]

Take a look at this thread to see how you can get a freshly new (non-corrupt) apache2 ispcp.conf file (requires scripting, it would be great if you contribute it)...

Obviously, you have to trigger a configuration regeneration afterwards