From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mike Christie Subject: Re: [PATCH] scsi_transport_fc: handle transient error on multipath environment Date: Fri, 12 Feb 2010 11:46:28 -0600 Message-ID: <4B7593F4.2050102@cs.wisc.edu> References: <4B750CB7.4030805@jp.fujitsu.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-2022-JP Content-Transfer-Encoding: 7bit Return-path: Received: from sabe.cs.wisc.edu ([128.105.6.20]:57461 "EHLO sabe.cs.wisc.edu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755059Ab0BLRqr (ORCPT ); Fri, 12 Feb 2010 12:46:47 -0500 In-Reply-To: <4B750CB7.4030805@jp.fujitsu.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Tomohiro Kusumi Cc: linux-scsi@vger.kernel.org, James.Bottomley@suse.de, linux-kernel@vger.kernel.org On 02/12/2010 02:09 AM, Tomohiro Kusumi wrote: > @@ -1953,6 +1987,13 @@ > { > struct fc_rport *rport = starget_to_rport(scsi_target(scmd->device)); > > + if (rport->recover_transient_error) { > + fc_queue_work(scmd->device->host,&rport->rport_te_work); > + scmd->result = ((scmd->result& 0xFF00FFFF) | > + (DID_TRANSPORT_DISRUPTED<< 16)); > + return BLK_EH_HANDLED; > + } > + > if (rport->port_state == FC_PORTSTATE_BLOCKED) > return BLK_EH_RESET_TIMER; - For the link down case you mentioned, would we see that the rport is blocked here then we would return RESET_TIMER. If the fast_io_fail tmo is set, then that would fail io quickly upwards (the fast io fail timo would probably fire before the cmd even timed out). What transport problems are you seeing where the rport is not blocked and the scsi cmd timer fires? Would it be mostly buggy switches or something like that? - Maybe you want to instead hook something into the dm-mutlipath's request (no more bios like in 2004 :)). Can you set a timer on that level of request. If that times out then, dm-multipath could do something like call blk_abort_queue. I think the problem with blk_abort_queue might be that it stops all IO to the entire host where you probably just want to work on the remote port/path. For that you could call something like recover_transient_error. Maybe it could just be a call to terminate_rport_io from a workqueue though.