Having OpenKM store files using file system

[phillman5]

We recently downloaded the community 3.0 version of OpenKM. In another post it was stated

\"Files on OpenKM can be stored on several ways, dbms ( postgre, mysql, oracle ) or file system.\"

We would really like to try just storing the files on the file system. Is there explicit directions on doing this?

[pavila]

You have to modify the repository.xml file to use a different backend. See http://jackrabbit.apache.org/ for more info because it is not simple to manage this. Or search in this forum because I remember some related posts.

[wiseman]

Can you provide a specific link, or better instructions so that we can have a better guide on how to do this? I am sure there are many out there who would like the backend to be a filesystem instead of a database, and detailed instructions would be great. If you provide the basic step by step guide, I\'ll convert it into a more detailed one, well translated with screenshots.

We MUST improve documentation.

[phillman5]

Thanks for the offer Wiseman! We have looked at several documentation control systems. OpenKM is the simplest of the bunch and I think we would use it. It does what we need with some nice extra features, quick learning curve and not get in the way, the only mark against it is the database repository and not the file system for the actual folder/files. So I ask again, if the developers could give explicit directions on a web page instead of telling us to go search that would be very useful, we are not programers, but at least technically literate and willing to try given directions.

[pavila]

You should read Support > Documentation > Install Guide. It is simple, but require some knowledge bout JDBC and Jackrabbit.

[alloydog]

We too would prefer to use a file system for the storage. It might not hurt if we use this thread to keep a track of what what we do and what happens.

I did a forum search for \"repository.xml\".
it turns up a few useful posts, such as:

Where are the files?

and

Repository (Spanish - I ran it through Google translate)

[phillman5]

Here is what one of the moderators (pavila) posted

If you want to know more about backend storage, you should read these pages:

* Jackrabbit Configuration http://jackrabbit.apache.org/jackrabbit ... ation.html
* Persistence Manager (PM) FAQ http://wiki.apache.org/jackrabbit/PersistenceManagerFAQ
* DataStore http://wiki.apache.org/jackrabbit/DataStore

It seems to me if they advertise that OpenKM can use the filesystem as the repository, they would give step-by-step instructions, not lead us to several that we have to digest in total to figure out how to do it.

[phillman5]

Wiseman
Are you still reading this forum? Did you see the three links Pavila left for using backend storage?

[phillman5]

I\'ve had a few off line e-mails with Josef. I was not phrasing the question correctly. The synopsis is:

Question: What we want is to set up OpenKM to put/store the repository folders/files into OS\'s folders/files, not a separate database. I.E. you would not need to export the repository. So even if OpenKM/mySQL/JackRabbit is not running (or has crashed), the administrator on the server could see or copy repository folders/files just using the OS. OpenKM would actually store folders/files in regular OS folders/files.

Answer: Let\'s setting some preliminars:
1- Jackrabbit default OpenKM configuration saves document and folders on your server disk, but it\'s not usable as you wish because uses a strange filename folders structure ( take a look under \"repository\" folder ).
2- Webdav support enables to see this repository ( saved on your server disk ) as a normal folder/document names, but this is not what you want, you\'re looking it for some backup purpose, webdav only it\'ll run well if repository is not corrupted. (Note you need the Enterprise addition for WebDAV support.)

You want a repository on your server disk with folders and documents with the same structure you view across OpenKM UI ( regular OS folders / documents ). We\'ll the answers in this case is that YOU CAN NOT DO IT. Why ? because jackrabbit uses a different structure -> one reason to understand it is that documents has several versions and it can not be set as a normal folder/document structure, because one document name has assigned several documents at the same time with the same name ( not possible on normal file system ).

If you\'re worried with losing information you must ensure are doing well backup. You must copy \"repository\" folder, and \"server/default/data\" folder -> try to copy it, and reinstall OpenKM with the backup on other computer or folder to test your backup goes right.

The last suggestion I made to you is, for example, connect a external hard disk to your server ( USB ) and run the export utility on administration -> this it\'ll save the same folder / document structure there ( only one problem you\'ll lose historical, security grants ). It\'s not the right way to doing backups.

[pavila]

Also note that you have to stop OpenKM (JBoss) before the backup of the \"repository\" folder to prevent modifications across the backup process.