The embedded database openfire uses is Hsqldb, written in Java. Openfire has a sort of migration-guide, but it is not exactly complete. First of all the program mentioned there, hsqldb-transfer, is:

  1. A GUI program...
  2. That must be run as the same user as the owner of the hsqldb database. Read and write-access is insufficient?!

If openfire is running on a server with no graphics, this means:

  1. Turning off the server on machine one
  2. Copying Path-To/embedded-db over an (offline) openfire-installation on machine two, which happens to also have graphics and the same openfire-setup as machine one (same plugins, version etc.)
  3. Starting openfire on machine two
  4. Restarting the setup-wizard so that the copied database is recognized
  5. Shutting off openfire on machine two
  6. "Upgrading" the user that runs openfire on machine two to be a fully fledged human user
  7. Logging in as the openfire user
  8. ... but since openfire is off, the database is now not on disk... which means that
  9. Running hsqldb-transfer to transfer the database

does not work.

In the guide the url jdbc:hsqldb:Path-To/embedded-db/openfire is used. Problem is: sometimes that file is there, other times it isn't. In my case: it was there on Monday when I did a dry-run without turning off openfire on machine one, it was missing today, when I did turn off openfire.

(The next steps in my successfull dry run was:

  1. While transferring, change datatypes not supported in the new server on the fly, as hsqldb-transfer itself is incapable of mapping between sql dialects
  2. Manually run the database-alterations on the external sql-server to get to the desired version (easy, since the existing version is in the table version). I went from 3.5.2 to 3.6.4...
  3. Dump the contents of that database and do an import on super sql server machine three, which is shiny and new
  4. Set up openfire on machine four, which is not so old that it's falling to pieces, to use the external database on three
  5. Profit!


How does one get a hold of an offline, not being updated copy of the database such that hsqldb-transfer can use it?


This is kind of a lame way to do it, but if you're having problems finding the file when it's offline (which I can't explain):

You can take a backup of a live hsqldb as long as a checkpoint has not occurred during the backup. A checkpoint will occur every time the log file fills up. There is a setting in the embedded-db's properties file:


(which is in meg - 50M)

As long as the log file doesn't hit that size during the file copy of the live db file, the file will be consistent. Whatever is in the log file has not yet been written to the db. If you can do this during off-hours, you can get a reasonably good dump file.

So.. you could kill any active sessions, do the file copy, shutdown the server, and then proceed from there.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.