Scalix Forums

I have a working Scalix (v. 11.4.0.11669) installation on Fedora 7 that i want to move and upgrade to CentOS 5.2 and Scalix 11.4.1

As described in several tutorials and threads the only thing you need to do is copy the /var/opt/scalix folder to the new machine. The Scalix installer will detect the old data and reuse settings and data.

- Both servers have the same hostname and domain (scalix abbreviation = dd)
- Both servers have the same IP. Changed the IP of the old server after stopping all relevant services. Just too make it possible to copy data between both servers.

1. CentOS installation

- Textual installation. The only selected packages are "Administration Tools" and "Base"
- Firewall enabled ports: ssh:22:tcp, http:80:tcp, https:443:tcp, smtp:25:tcp, openmail:5729:tcp imap:143:tcp
- SELINUX = disabled

After first login


2. Disable original server

Determine scalix group and userid (both are set to 499)


3. Prepare new server

Create scalix user

Synchronize /var/opt/scalix


4. Install Scalix

Installation steps

- Accept license agreement
- Run unpacked package automatically
- Install all scalix components
- System check gives OK on all items
- Installation ends with a fatal error

Fatal error

Add /opt/scalix/bin to PATH


5. Health check + fixes

The script fixes some permission issues. When i run omcheck again the output confirms that the problems are solved by running the output script.

6. Installation log


7. Problems & troubleshooting
I've tried a clean installation without importing old data on the same server setup. That works just fine.

Tried http://www.scalix.com/forums/viewtopic.php?t=6219
pgstartup.log

Line 104 in postgresql.conf



outputs


Tried http://www.scalix.com/forums/viewtopic.php?t=6920&postdays=0&postorder=asc&start=0






8. Conclusion
I think i'm facing two problems

1. The version inconsistency
2. The failure to set the password for the database user 'scalix'[/code]

"ompatchom" will fix the version inconsistency.

I'd rerun the installer, and delete the database, then make sure /var/opt/scalix/dd/postgres is deleted, and then I'd run the installer again, and reinstall the database, which will also reconfigure it. I'd bet that the version of Postgres has changed, and there's no data in there that matters, so nuking it and recreating is easier than troubleshooting it through an upgrade.

Note that once postgres is reinstalled, it will take a (very) long time to regenerate the data, and this will likely be noticeable performance wise. Not the end of the world, but be aware of it.

BTW, VERY good decision to move to Cent. VERY good. Scalix does not belong on Fedora.

Kev.

1. Version
ompatchom returns


A start or a stop of the scalix daemon (/etc/init.d/scalix) returns no result and does nothing.

2. Database
Noticed something unusual before trying to remove and reinstall the database. I was looking for the postgres folder in i found it under



Why de double 'dd' folder ?

I uninstalled the database (option 4 in the (un)installer). Also confirmed the deletion of the cached data.



The postgres folder still exists in /var/opt/scalix/dd/s/dd/

The installer detects that the DB is removed because i was able to reinstall the database




Returns no result

P.S.: I rebooted the machine before trying to fix the version number and database problem

This is the mess I was referring to in another thread. :)

Ok, stop Scalix.
omshut -t3
/etc/init.d/scalix-tomcat stop
/etc/init.d/scalix-postgres stop

mv everything from /var/opt/scalix/dd/s/dd up 2 levels so that it's at /var/opt/scalix/dd, and then remove the (now unused) directory.
mv /var/opt/scalix/dd /var/opt/scalix/dd_old
mv /var/opt/scalix/dd_old/s/dd /var/opt/scalix

This should leave you with /var/opt/scalix/dd/postgres (for example). And there should no longer be a /var/opt/scalix/dd/s/dd directory.

THEN...

Check /etc/opt/scalix/instance.cfg. Verify that omdatadir is /var/opt/scalix/dd/s. If it isn't, then change it to be that.

This is exactly the same as the file /opt/scalix/global/config. Make sure the datadir is correct there too.

Then rerun ompatchom. Now it should work. Then uninstall/reinstall postgres, and you're golden. I suspect, if you used rsync to move this, that it was wrong on the original server, and we're actually cleaning that up on the new server. So be aware that if you rerun the rsync, it'll overwrite this all unless you change the config a bit.

Kev.

It worked. Both version and postgres problems are fixed. I had to reinstall tomcat also, otherwise webmail and sac are not available.

Weird thing is that there was no double 'dd' folder on the old Fedora 7. Probably it was created by the installer when i ran it the first time

Thanks for the help !

If it wasn't there on the old version, then I'd bet that either the rsync had some misspecified parameters, or the installer tried to reupgrade the messagestore to a new location. In either case, it something that has been seen lots of times before, and it's an easy fix.

Either way, You're WAY better off on Cent. Enjoy.

Kev.

Finally after testing on fc7 I moved my scalix installation to cent os 5.2

Facing the same problems like wittewim this thread helped me a lot to get the move done more or less.

sac and webmail are still not avaiable, but I am not able to uninstall/reinstall tomcat as wittewim did. When I choose tomcat and or the tomcat apache connector for uninstallation in the install dialog the entry will not accepted.

Did I missed something to reinstall tomcat ?

This may help to get sac and webmail working again.

First look for the following symlink in /etc/httpd/conf.d


If you don't have it, type the following command:


Then edit /etc/httpd/conf.d/scalix-tomcat-connector.conf and change the Include line that has 'apps' in it to 'instance', like this:


Restart apache and scalix-tomcat and test.

HTH,
Lance