[OSM-dev] Eating all available ram on www

Nick Hill nick at nickhill.co.uk
Wed May 9 01:11:28 BST 2007


Hi All

Looks like we have a teething problem on the rails port where the RAM footprint 
grows over time, eventually eating all, then the process closes.

We had a similar issue with the previous API which was covered over by setting 
apache to restart the process every 25 requests.

I guess it is an issue of the garbage collector not freeing memory from unused 
objects. Perhaps there is a form of referencing such that the GC doesn't know 
the object is no longer used so doesn't 'know' to free the memory, in which 
case, some explicit freeing may be needed.

A patch for the rails port or a knowledgeable specific explanation why Ruby/API 
memory footprint keeps growing would help a lot.









More information about the dev mailing list