[mvapich-discuss] IBV_EVENT_QP_LAST_WQE_REACHED error

[Matthew Koop]

Just to update the list -- this has been tracked down to system calls in
the application and an old kernel that does not have the fork support for
OpenFabrics.

Matt

On Wed, 7 May 2008, Matthew Koop wrote:

> Bill,
>
> The error described in the link you gave was the result of some 'fork's or
> system calls in the code (that did a fork). Does your code contain any
> system calls or forks? If so, please look at section 7.1.2 of the user
> guide:
>
> https://mvapich.cse.ohio-state.edu/support/mvapich_user_guide.html
>
> Also, you may want to consider upgrading to OFED 1.3 and MVAPICH 1.0,
> which has additional scalability and stability enhancements.
>
> Let us know how things go. Thanks,
>
> Matt
>
>
>
> On Tue, 6 May 2008, Bill Riner wrote:
>
> > We upgraded from IBGold to OFED 1.2.5.5 and we also get the "Got FATAL event
> > IBV_EVENT_QP_LAST_WQE_REACHED" error that's discussed in this thread:
> >
> >
> >
> > http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/2007-October/001139
> > .html
> >
> >
> >
> > [0:compute-0-23.local] Abort: [0] Got FATAL event
> > IBV_EVENT_QP_LAST_WQE_REACHED, code=16
> >
> >  at line 2555 in file viacheck.c
> >
> >
> >
> > The particular code worked fine before the upgrade.  The error doesn't occur
> > when using just one processor.  MPICH works fine on any number of
> > processors.   The OSU and Intel MPI benchmarks work fine.  Other MPI codes
> > work fine also.  We're using MVAPICH 0.9.9 that came with OFED.
> >
> >
> >
> > Has this been resolved?
> >
> >
> >
> >
> >
> >
>
> _______________________________________________
> mvapich-discuss mailing list
> mvapich-discuss at cse.ohio-state.edu
> http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
>