Hi list, Sorry for wrong patch submitting of the previous email. I rewrite the email correctly here: I reviewed all the rlm_sql_* modules hunting for possible other bug and I think I found another one. This time I cannot replicate an environment for testing, so I made a patch and atteched here instead of opening a new bug. The bug is locate in rlm_sql_oracle module and it involves the same config->sql_db as the rlm_sql_iodbc one. Thank you for your time! --- sql_oracle.c.orig 2010-09-28 13:03:56.000000000 +0200 +++ sql_oracle.c 2011-06-13 19:15:32.816405400 +0200 @@ -176,7 +176,7 @@ if (OCILogon(oracle_sock->env, oracle_sock->errHandle, &oracle_sock->conn, config->sql_login, strlen(config->sql_login), config->sql_password, strlen(config->sql_password), - config->sql_db, strlen(config->sql_db))) + config->sql_server, strlen(config->sql_server))) { radlog(L_ERR,"rlm_sql_oracle: Oracle logon failed: '%s'", sql_error(sqlsocket, config)); sql_close(sqlsocket,config);