Memory Issue using OO (soffice) as batch conversion service

Question

I have a Java Web Application that uses an OOo instance (on the same environment) started in headless mode as batch converter.

• environment is RHEL 5.5.
• The OpenOffice version is 3.1.

I've run some tests and I found that after about 1000 conversion the resident memory of the soffice process was around 99% of the total. OpenOffice finally crashed.

I would use this server as a batch converter for documents, but this memory issue is requiring a daily restart according to the medium amount of conversion request that we have.

I know that this memory issue is affecting OO since release 2. I would know if there is a solution, or if using a Windows Server instead of Linux could solve the problem.

Thank you.

Answer 1

You may be interested in "OpenOffice Server Daemon".

By managing the conversion it enables you to:

• take advantages of multithreading during conversion by parallelizing them
• reduce the impact of memory leaks by restarting each soffice process in case of a crash and/or after a predefined number of conversions

A good point is you don't have to modify any of your code, the interface to the server is the same than the one to OO.

Answer 2

I currently have a similar problem. I don't have the possibility to switch to a Windows host for OOo in my environment so I'll have to look for a solution for Linux.

As it seems you have the possibility to switch I can tell that on my workstation (a Windows 7) the consumption is there but only about 15% of that of the linux machine (but there is still a memory loss per converted file).

If I find the way where it does not happen at all, I'll let you know. (If you only do conversion you might want to look at: Jodconverter. As has been suggested here. )

EDIT: I found this oooServer to handle multiple instances as well.