On Mon, 15 Sep 2008, [EMAIL PROTECTED] wrote: | Hi all, | | I observe such problem spawning randomly on my nodes: | | kernel: ib_ipath 0000:03:00.0: RXE parity, Eager TID port 0 idx 0x33c expected 20447819, but got 20047819. | kernel: ib_ipath 0000:03:00.0: infinipath0: RXE parity Eager TID not recoverable, read 20047819, expected 20447819 | kernel: ib_ipath 0000:03:00.0: infinipath0: RXE parity, Eager TID error is not recoverable | | That's for qlogic cards, Mellanox ones seem to be much, much more stable. As I need to stress every card I just looking for a tool to make memory chips there under heavy load and, unfortunately, with not much luck. So what's the tool for diagnosing the cards?
There is no memory on the card, this is on-chip memory. The only test tool for it is a QLogic internal manufacturing test tool. If you are seeing this more than once on the same card, you should get the card replaced by contacting QLogic support. Some memory errors are inevitable. We try to recover from them, but not all of them are recoverable (have a "known good" backup, or are known to be safe to rewrite and continue). Dave Olson [EMAIL PROTECTED] _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
