Problems after upgrading from 2.0 to 2.2

[Josep]

Today I decided to bite the bullet and upgrade my server from 2.0 to 2.2 and boy, what a disaster!
I had cleaned up everything prior to the upgrade and we run backups everyday (including configuration).
We don't run many services, so I thought it should be pretty straightforward.
I couldn't be any more wrong.

Downloaded the migration tool and run it.
At some point it claimed some dependency problems related to squid were preventing it from completing.
I promptly uninstalled the squid package on another session, but eventually I had to terminate the original script.
These are some of the last lines:

Code: [Select]

No apport report written because MaxReports is reached already
                                                              Processing triggers for libc-bin ...
ldconfig deferred processing now taking place
Errors were encountered while processing:
 zentyal-core
 zentyal-objects
 zentyal-services
 zentyal-network
 zentyal-firewall
 zentyal-antivirus
 zentyal-ca
 zentyal-dhcp
 zentyal-dns
 zentyal-ebackup
 zentyal-users
 zentyal-ftp
 zentyal-monitor
 zentyal-openvpn
 zentyal-samba
 zentyal-software
 zentyal-squid
E: Sub-process /usr/bin/dpkg returned an error code (1)
Command FAILED! Please check your internet connectivity

Press return to continue or Control+C to abort...

After that, I cleaned up and everything was properly installed.
However, the system was left unconfigured.
Backup configuration, FTP configuration, Users, etc. everything is lost.

Trying to recover from the 2.0 configuration backup does not work, because, you know, who would ever need it? right? Sorry about my rant but at this point I'm really pissed off at how  poorly handled  the migration process is.

I haven't found many migration-related posts, so either I'm in a situation of really bad luck, or I just don't know.

Does anyone have a quick solution for this? or will I have to actually go and recreate all my certificates and users from scratch?

2012-02-21 --Update--
In the end I had to remove all ebox and zentyal packages and start from scratch.
I have managed to salvage data from the configuration backup and I have been able to restore users and groups information directly into the LDAP database.
I also have been able to rebuild all my shares and their permissions by carefully reading the Redis files.
No luck with certificates, I didn't have any more time for this. I will re-issue new certificates to everyone.

Too bad that something that could be accomplished in under 1 hour ended up taking 12 in a small organization.

[Lueghi]

I'm standing exactly at the same point!
But I'm not so familiar with linux to know all the things for finishing the upgrade. I did 2 times a new install in the last year for our SOHO-system and I don't want to do it again.

So can anybody show me the way ahead?

[Josep]

Lueghi, from my experience, there is no guarantee that any path will work.
I waited patiently for months that the new version was ironed out and when I thought it would be safe I upgraded and ended up in the mess that I reported.

It is my opinion that if Zentyal 2.0 serves your needs, there is no real reason to upgrade. After all, in a few months you will have Zentyal 3.0 and will have to start it all over again. Considering the amount of time I had to devote to solve all the issues that arose from the failed upgrade, it would have been cheaper to just buy Windows Server and get done with it, wasn't it my own bet in this technology.

If you go ahead, make sure you have a full backup of everything on an external device. Make configuration backups to folders on the server and to your own computer. Make a copy of the LDAP contents in LDIF format (this may allow you to make adjustments in plain text if necessary). Purge (not just remove) anything that you are not actually using, as previously installed packages and configurations may interfere with the upgrade.

Finally, use Clonezilla or some other software to actually mirror your disk, as it may be a lot faster to go back should something go terribly wrong.

Good luck.

[Lueghi]

Hi Josep,

thanks a lot for your answer ... but you can imagine that I'm not happy now (it's not because of you !!).

I waited patiently for months that the new version was ironed out and when I thought it would be safe I upgraded and ended up in the mess that I reported.

This was exactly my intension too.

I did the update because I assumed that there will be no direct update from 2.0 -> 3.0.

The server is still running and I didn't miss a function until now except the zentyal GUI. It seems the port has changed, apache itself is running. But I don't know what the server will do if I have to do a reboot .....

So I will forget about this until I get real problems  . When this happens I have to decide what do to   .

[jmcm]

I'm also in the same boat, the update started, the modules were deleted, and then began the downloads of the new ones, until after some time the downloads began to fail due to loss of internet connection, and the upgrade stops there. If I retry, I've no success, if I stop the script and try to run it again, then there is no module to upgrade, because they all were deleted before ...
Why start deleting things before being sure that everything is available and transferred to the local disk?

At least I had a clone of the disk, so it was easy to go back to version 2.0

I assumed that it was just bad luck with the loss of Internet connection, or something I had done wrong, so I checked everything and tried a second time with the same result, so for now I'll stick with the version 2.0.

[Lueghi]

Hi Josep,

thanks a lot for your answer ... but you can imagine that I'm not happy now (it's not because of you !!).

I waited patiently for months that the new version was ironed out and when I thought it would be safe I upgraded and ended up in the mess that I reported.

This was exactly my intension too.

I did the update because I assumed that there will be no direct update from 2.0 -> 3.0.

The server is still running and I didn't miss a function until now except the zentyal GUI. It seems the port has changed, apache itself is running. But I don't know what the server will do if I have to do a reboot .....

So I will forget about this until I get real problems  . When this happens I have to decide what do to   .

Only for information:
my server is runnung since 286 days without any problems in the corrupted installation ... so why should I change something

[Lueghi]

I would like to reactivate this threat .... because I have a little problem now:

My old server ist running now with an uptime of 495 days without any problem in face of the corrupted installation. But there was no need to do some modifications. Now I want to switch step by step to a new installation with zentyal 3. For the fist step I wanted to move the services dns and dhcp to the new zentyal server. But I'm not able to stop the services on the old machine. Everytime I stop the service or kill the process the service is restarting automatically . So my question is: what is the way to stop the services without shuting down the old server. My fear is that after a shutdown the old server will not run again in the way he is running now because of the corrupted zentyal installation.

Thanks for your help.

[Sam Graf]

So deactivating the DNS and DHCP modules running on the old server is not working? If that's the case, I would think you would have to delete the components.

I don't know of any way to permanently disable a module without modifying the server's configuration in some way. That's short of a restart, but it still requires modifying the configuration. And I'm not smart enough to know at what point in a reconfiguration process you might meet the corrupted installation.

Hopefully you will back up any crucial data before doing anything ...

[Lueghi]

So deactivating the DNS and DHCP modules running on the old server is not working? If that's the case, I would think you would have to delete the components.

I don't know of any way to permanently disable a module without modifying the server's configuration in some way. That's short of a restart, but it still requires modifying the configuration. And I'm not smart enough to know at what point in a reconfiguration process you might meet the corrupted installation.

Hopefully you will back up any crucial data before doing anything ...

You wrote deactivating the modules. I tried to stop the services with the init-scripts. Do we mean both the same or is there a way to DEACTIVATE the services from the console?

[Sam Graf]

I meant from the GUI. Uncheck the module. That should shut down the service.

[Lueghi]

I meant from the GUI. Uncheck the module. That should shut down the service.

The GUI isn't available since my experiment of installing Zentyal 2.2 ....

[Sam Graf]

Then I would move to the next step--remove the modules. You'd have to do that manually if software management also is broken.

The only other thought I have, one that's less invasive, would be to try to block services via the firewall. Since I've never tried to do what you're trying to do I'm just thinking out loud, but if you can prevent the old server from responding to DNS and DHCP requests (and so on), that may be get the job done ... ?

[Lueghi]

When I understand zentyal right, then this is a GUI for configuring an ubuntu installation. Zentyal will write all the final configuration files into the standard config directory /etc. So I will try to move the config-files for dns and dhcp out of this directory and will then kill the services. I will tell you my experience ....

[half_life]

You could just remove the executable bit from /sbin/dhcpcd3 then try and stop the service.  Caution, either action might create a race condition since it seems to be auto restarting.

[Lueghi]

You could just remove the executable bit from /sbin/dhcpcd3 then try and stop the service.  Caution, either action might create a race condition since it seems to be auto restarting.

I did it exactly in this way ... and it worked . The migration phase has startet .

Thx for your help.