Le vendredi 25 mars 2011 à 12:30 +0100, Yann Droneaud a écrit :
> But the only event I caught and handle is IBV_EVENT_COMM_EST because
> it happen sometimes under high load (in this case, rdma_notify() is
> used to send the event back to RDMA CM layer). 

Oh, but it doesn't work if the server (accepting connection) is really
too slow (like running under strace), I'm getting error when calling
rdma_notify() : it returns -1 and errno is set to EISCONN : Transport
endpoint is already connected.

And then the communication process between the two peers is blocked.

(BTW, not using rdma_notify() doesn't fix this problem, so there's
another problem.)

My question is still valid: should I monitor IB async event when using
RDMA CM ?

-- 
Yann Droneaud
OPTEYA





_______________________________________________
ewg mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg

Reply via email to