Guess what… everything was reset!!! The users, I even couldn’t enter with the admin user (not even with the default password)…
That does not sound like an over-write by the installer. That is a corruption of the embedded database file by something happening on your server.
A good installer should have only replaced the core of openfire server and maintained the user db, settings, etc.
A good suggestion. This is a community of people serving others for voluntary reasons. You are most welcome to join the effort and improve the installer.
I’m afraid improving the installer can be tricky, as we use install4j commercial installer and i don’t know what options do we have to customize it. Probably there is some “rules” files for it to build installers for Openfire and Spark. But i don’t have access to Bamboo to check it (our build system).
Nuno, you said, that installer has removed all of your files. Are you sure? Were you looking at openfire folder while doing the upgrade? I agree with Dele that this looks more like a corruption as you would get a new setup process upon starting Admin Console if everything was wiped.
I have gone through the Openfire build ANT script as well as the Install4j install scripts and there is no reference to the embedded database folder or files.
I then looked the Openfire source code and found that the creation of the folder and file is done by Openfire when the user runs the setup wizard the first time.
File databaseDir = new File(JiveGlobals.getHomeDirectory(), File.separator + “embedded-db”);
// If the database doesn’t exist, create it.
if (!databaseDir.exists()) {
databaseDir.mkdirs();
}
Conclusion: The Openfire 3.7.1 upgrade could not have deleted or corrupted the embedded database as it is unaware of it. In fact, I bet that if you manually uninstall Openfire, the embedded database folder will be left behind as Install4J did not create it in the first place.
Another thing. If Openfire starts with out embedded-db folder and settings it shows Listening at http://127.0.0.1:9090. If it was showing http://yourdomain:9090, then it was a corruption of some sort and not the absence of db and settings.
I after reading all the posts on how to do it after reading everything went fine for most, I decided to upgrade last night.
We also use shadow copy, but for best practise, I decided to close down spark (along with the service), take a copy of the entire openfire folders (located in program files x86), then run the upgrade.
The upgrade went fine and straight away launched me into the new admin interface.
A quick look around and everything looked good. I logged off feeling happy with the upgrade.
I came into work this morning to find openfire not running, and when manually launched (I’d previously setup the service for it), it claims that the DB was already in use by another process.
It didnt matter what I did with the software/service, I could not get into it.
I decided to reboot the server and it seems to have sorted itself out, everything is up and running.
For information, Im using the desktop database, not SQL.