This looks suspiciously like a problem I have had before, where somebody writes a script to delete files that look like temp files, no file extensions, over a month old.  I had one that was deleting classes created at runtime, so each morning there was a good chance of getting classloader errors.

Best of luck.

From:        "Shah, Sumit (CGI Federal)" <>
To:        "" <>
Date:        20/12/2011 11:58 AM
Subject:        Jackrabbit 2.2.5 - loss of data

Hi All,

I am running into a serious issue. It seems like I am unable to retrieve documents from Jackrabbit that are more than a month old. I get the following error:

"JCR Action 'Get stream' cannot be performed because the provided path does not exist"

I am running Jackrabbit in standalone mode and also in a clustered environment. I am seeing the same issue on both. When does this happen? Is there a self initiated process that cleans up the data within Jackrabbit? What are the possible resolutions to this?

I would appreciate any help on this.