Hi Eric - Can you send me (offline) the build log so we can start there ?
I dont think what Becky mentioned is going to cause this issue since its buried very deep in IB code, but you definitely will want to look at that too. Thanks, Kyle Schochenmaier On Fri, Mar 26, 2010 at 9:16 AM, Becky Ligon <[email protected]> wrote: > Just FYI: Centos 5.3 does not deliver the latest version of Berkeley DB. > I strongly suggest that you go to the Oracle site and download the latest > version. We have experienced DB corruption problems here, seemingly from > thread issues, that have gone away after using the latest version of BDB. > We are running Centos 5.4 x86_64. > > Becky > -- > Becky Ligon > PVFS Developer > Clemson University > 864-656-3865 > >> >> >> Kyle, >> >> Thanks for responding. All of the following work correctly between my >> I/O servers: >> >> ibv_uc_pingpong, ibv_rc_pingpong, ibv_srq_pingpong, ibv_ud_pingpong, >> ib_ping >> >>> Also, did everything work both prior to the hardware change and >>> software (ofed) change ? >> >> We upgraded our I/O servers from >> Platform OCS / RHEL 4 >> OFED 1.1 >> Mellanox Technologies MT25204 [InfiniHost III Lx HCA] (rev 20) >> >> to >> >> Centos 5.3 >> Qlogic's version of OFED 1.4.2. >> Qlogic InfiniPath_QLE7340 >> >> We had no problems before with the old set up and we have our >> pvfs volume mounted over ethernet right now with no apparent problems. >> >> Thanks! >> >> Eric >> >> >> >> >> On 03/25/2010 10:25 PM, Kyle Schochenmaier wrote: >>> Hi Eric - >>> >>> Its been a while since I've looked at the code here but when we have a >>> failure in this location its usually due to an API change or a basic >>> configuration error. >>> >>> Can you confirm that the hardware is working via some standard >>> ibv_pingpong utilities? >>> >>> Also, did everything work both prior to the hardware change and >>> software (ofed) change ? >>> >>> >>> I'll try to see what else has changed if it is an API issue. >>> >>> Best, >>> >>> Kyle Schochenmaier >>> >>> >>> >>> On Thu, Mar 25, 2010 at 2:16 PM, Eric J. Walter <[email protected]> wrote: >>> >>>> Hi, >>>> >>>> I have recently upgraded the card in our pvfs I/O servers to the >>>> Qlogic QLE7340 (with hca_id = qib0) using Qlogic's version of the >>>> ofed 1.4.2 drivers. I get the following message server logs once I >>>> tried to start a second server: >>>> >>>> Error: init_connection_modify_qp: ibv_modify_qp RTR -> RTS: Invalid >>>> argument. >>>> >>>> all servers then die. Running with all messages logged showed nothing >>>> extra. >>>> >>>> I have tried both versions 2.8.2 and 2.8.1 of pvfs, and I have tried >>>> the stock ofed 1.5 as well. All combinations give this problem. >>>> >>>> Is there anything I can do to solve this problem? >>>> >>>> Thanks in advance for any help. >>>> >>>> Regards, >>>> >>>> Eric J. Walter >>>> Department of Physics >>>> College of William and Mary >>>> >>>> >>>> >>>> _______________________________________________ >>>> Pvfs2-users mailing list >>>> [email protected] >>>> http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users >>>> >>>> >> >> _______________________________________________ >> Pvfs2-users mailing list >> [email protected] >> http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users >> > > _______________________________________________ Pvfs2-users mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
