Cannot restart install

[Shaun]

Hi everyone.

I'm trying to upgrade a 11.04 install to 11.2 - the upgrade has to be aborted (Looks like a DNS issue, but I'm working on that but my problem is that if I abort, the install script won't try again, as it thinks that 11.2 is installed. I've tried recovering the /var/opt/scaix, opt/scalix and /etc folders, but I'm obviously missing something. Can anyone tell me how the installer works out which version of Scalix you're running, and how I can correct the situation so the installer can re-run?

[schmoe90]

Just run the scalix-installer, remove the components then run it again and install the components.

[Shaun]

Ok - thanks for that, I'll try it.

The question however, still stands. I have to do this upgrade remotely, and I need a way to recover the entire config if it all goes wrong. My current strategy is to backup /var/opt/scalix, /opt/scalix & /etc, but this isn't enough. Scalix must record the details of the packages installed somewhere else - can anyone please tell me what other directories are involved? I need a way to regress the ENTIRE scalix install - not just the data.

[schmoe90]

If you need to regress the entire install, you'd basically delete it and reinstall it at a lower version.

[Shaun]

Mnnn...

I don't really want to remove it.

What I would like to do is the following.

1. backup the entire working scalix install - everything, data, programs, config, the lot.
2. try and run the upgrade to completion on my existing machine
3. if the upgrade fails, recover scalix back to how it was prior to the upgrade attempt.

If this was local (i.e. in my office), I would back up the machine using Paragon Backup, try the install, and if it didn't work, recover the disk using Paragon.

I don't have that luxury, as the machine I want to upgrade is the other side of the world to me.

I've already tried rerunning the install (fails, as it thinks 11.2 is already installed - it isn't)
If I remove the scalix components, and then recover back my tar backups, postgres doesn't start - which would indicate that directories other than those I mention previously are involved. Trying to regress the server back by using the install program might work, but what if that fails? I'm left in limbo.

All I need is a list of directories that scalix uses in normal operation, and during install. With that, I can tar up the directories and recover the system in case of problems.

[Shaun]

Ok - found an answer. You need to backup the following directories. If you recover these, it seems you can regress scalix..

/var/opt/scalix
/opt/scalix*
/etc
/var/lib/rpm
/usr/java
/usr/lib/jvm

This is everything, user data, the lot.

The reason install won't rerun is because the installer uses rpm to install the packages - the rpm database thought that 11.2 was installed, even though I had regressed everything else back.