today i am attending the FCP-4 meeting in the morning. we are spending time discussing what makes an I_T_L_Q nexus in fibre channel. the discussion revolves around some problems that are being experienced in the field during error recovery. so far there is some agreement around the fact that the Q portion of the nexus is a fully qualified combination of the OX_ID and RX_ID but there is belief that that is not enough. turns out if the target uses the S_ID and OX_ID combination then there should be sufficient information to fully identify the I/O process. however, during retries it is required to add the additional task retry identifier – if retires are enabled. Q is equal to OX_ID, RX_ID, and task retry identifier. otherwise a fully qualified OX_ID and RX_ID identifies the Q. how exciting is this.