All parallel sessions running on one node?
922140Mar 6 2012 — edited Mar 7 2012Gurus,
we got a full rack Exadata, 8 nodes.
we're facing a problem while executing parallel query/dml in a full Rack Exadata. If we issue a parallel query via hint or "alter session force parallel query", all PX processes just run on node 2. No matter how many times we retry, how many queries we issued, and no matter on which node we issue the query, all PX processes just run on node 2, but never on other node.
e.g, we issue "select /*parallel (t 16) */ * from t" on nodes 6 and 7 at same time, all more than 32 PX running on node 2, even none of them was on node 6 or 7.
We're setting parallel_degree_policy to MANUAL. All other parallel related parameters are all set as default.
Does anyone encounter similar problems? Any help will be appreciated.
Flo