On Wed, 2007-02-21 at 18:05, Sean Hefty wrote: > >>We haven't looked into this in more detail yet. This was our observation > >>while > >>testing on a larger (64 node) cluster this morning that we don't have > >>access to > >>at the moment. With the local SA cache running, we were surprised to see > >>any > >>retries, and when we looked into it more, retries were always for loopback > >>connections. > > Our investigation showed a couple of things. When we pulled our systems off > into a small cluster and ran opensm, things were fine. The cache was working > as > normal, and we did get loopback paths from opensm. > > On our development cluster, the cache was never getting any path records. It > would issue a GetTable query, and the SM would respond. The response had a > status of 0 (success), but never returned any path records. I believe that > the > SM node is running OFED 1.1.1.
I'm unaware of any changes in this area of OpenSM which would cause this but maybe I'm forgetting something. Can you run opensm with -V and send the logs to me ? This should be instructive. -- Hal > I don't have the ability to modify the kernel on the larger 64-node cluster > that > we were testing on to see what is going on there. > > - Sean _______________________________________________ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general