Best Practice For Taking Live Database Offline

[cmcfarling]

We use several databases hostsd on a FileMaker server for job tracking and billing. From time to time I need to make changes to the various databases. Minor changes can be made while users are logged into the db's. More involved changes generally require all users to be logged out of all the db's so I can get access to all the field definitions, relationships, etc. It's not really feasible to have everyone log out of all db's for an extended period of time during business hours. Due to that, most major changes have to be done on weekends.

To get around this, my thinking is to leave the live db's running on the server and copy the set of db's locally to a machine running FMP. I could then work on the duplicate set, making neccessary changes, for as long as it takes. When done i could then just move the altered db's back to FMP Server and all is well. Of course I would have to account for any data that was added/altered in the time it took to make my changes.

The big problem with this approach is the relationships configured in the databases. This sytem consists of 16 db's, most of which have muliple relationships set up. The relationships are all "hosts" based. In other words, when each relationship was created, the related file was chosen from the network hosts list. So if I copy one of the db's locally and open it, it automatically opens the live db from the server due to the relationship. To counteract this behavior, I turned off FMP's network protocol so I would be forced to choose a new database to assign to any relationships that are referenced. While this seems to work for some relationships, I still end up with several <File Missing> errors in the Relationships dialog box. Besides, there are so many relationship references and scripts that reference host based db's that I would never know if everything was correct.

So my question is this... what do FMP pros do in situations like this where there are live db's hosted on an FMP server that need major changes done to them? Is there a good way to make a duplicate set of the db's and easily alter the relationships in the duped set so that the relationships don't point to the live db's?

Thanks for any feedback.

Chris M

Version: Server v5

Platform: Windows 2000

[cmcfarling]

Well, I think I answered my own question. Initially, I had renamed the duplicate offline files. I put "_w" in their name to signify to me that they were working files. I tried again without renaming them and all is well.

[BobWeaver]

I've never run into any problems with the file references getting lost when moving a filemaker solution. If you select relative path and keep all the files together in the same folders (relatively speaking) there should be no problem. If you have used absolute paths for the file references, then it can be a problem.

If you do move the files to a different machine, I highly recommend that you unplug the network cable from it before firing up the database. Not only do you not want to accidentally access one of the server files while you are working, but you also do not want other users to accidentally access your copy of the files instead of the server copy.