On Wed, Sep 2, 2009 at 12:17 PM, Chris Worley<[email protected]> wrote: > On Wed, Sep 2, 2009 at 12:05 PM, Bart Van > Assche<[email protected]> wrote: >> On Wed, Sep 2, 2009 at 7:27 PM, Chris Worley <[email protected]> wrote: >>> >>> Then, the majority of messages were the pair: >>> >>> Sep 2 17:12:22 nameme kernel: [408676.552666] [0]: scst: >>> __scst_init_cmd:3361:Too many pending commands (50) in session, >>> returning BUSY to initiator "0x0002c903000260470002c90300026047" >>> Sep 2 17:12:22 nameme kernel: [408676.552670] [0]: scst: >>> scst_set_busy:366:Sending QUEUE_FULL status to initiator >>> 0x0002c903000260470002c90300026047 (cmds count 79, queue_type 1, >>> sess->init_phase 3) >>> >>> # grep -e "returning BUSY to initiator" /var/log/messages | wc -l >>> 221834 >>> # grep -e "Sending QUEUE_FULL status to initiator" /var/log/messages | wc -l >>> 167948 >>> >>> Maybe the initiator has stopped due to its "busy" tracking... deciding >>> the target cannot continue? >>> >>> Is the queue settable to larger than 50 outstanding commands? >> >> Definitely. The details are explained in the document srpt/README. > > This is looking god so far; I've set it to 256, and am not seeing any > queue-full messages. This may be the work-around.
No such luck; it takes much longer for the hang to occur, but it eventually occurs :( Thanks, Chris > > Thanks, > > Chris >> >> Bart. >> > _______________________________________________ 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
