Currently atlassian-crowd-openid-server.log and database directories are created in the directory where crowd is executed from. These files should be created in a known location in which the executing user has a read/write access.