From: kbusch@kernel.org (Keith Busch) Date: Wed, 24 Apr 2019 14:44:17 -0600 Subject: [PATCH v2 0/2] Adding per-controller timeout support to nvme In-Reply-To: <983e5d039dce9de1d32c71d28fd59bbc01c3fee5.camel@infradead.org> References: <20190403123506.122904-1-mheyne@amazon.de> <20190424200706.GB15412@localhost.localdomain> <983e5d039dce9de1d32c71d28fd59bbc01c3fee5.camel@infradead.org> Message-ID: <20190424204417.GC15412@localhost.localdomain> On Wed, Apr 24, 2019@10:30:08PM +0200, David Woodhouse wrote: > It isn't that the media is slow; the max timeout is based on the SLA > for certain classes of "fabric" outages. Linux copes *really* badly > with I/O errors, and if we can make the timeout last long enough to > cover the switch restart worst case, then users are a lot happier. Gotchya. So the default timeout is sufficient under normal operation, but temporary intermittent outages may exceed it. It'd be a real dissappointment if the command times out with an error anyway after waiting for the extended time, but we ought to cover the worst case time for a successful completion.