From: Fields Bruce <bfields@fieldses.org>
To: Trond Myklebust <trondmy@primarydata.com>
Cc: "linux-nfs@vger.kernel.org" <linux-nfs@vger.kernel.org>
Subject: Re: [PATCH 09/10] SUNRPC: Change TCP socket space reservation
Date: Wed, 6 Jul 2016 16:17:48 -0400 [thread overview]
Message-ID: <20160706201748.GG18856@fieldses.org> (raw)
In-Reply-To: <A6F54110-7D15-4545-B15B-D058EAB358F0@primarydata.com>
On Fri, Jun 24, 2016 at 09:31:07PM +0000, Trond Myklebust wrote:
>
> > On Jun 24, 2016, at 17:18, J. Bruce Fields <bfields@fieldses.org> wrote:
> >
> > On Fri, Jun 24, 2016 at 10:55:51AM -0400, Trond Myklebust wrote:
> >> Instead of trying (and failing) to predict how much writeable socket space
> >> will be available to the RPC call, just fall back to the simple model of
> >> deferring processing until the socket is uncongested.
> >
> > OK, it would be a relief to get rid of that, I guess that explains the
> > previous patch.
> >
> > But was there some specific reason you were running into this?
>
> I’ve been testing using a 40GigE network, which easily hits this bottleneck. The result is that each connection saturates long before we’re even near the saturation of the network or even the disk. So we find ourselves with a server that can continue to scale up to several 100 clients, but with each client seeing the same low performance irrespective of the total number of clients.
OK, I'm stealing some of your responses for a slightly more verbose
changelog (see below), but otherwise committing the patches unchanged
for 4.8. Thanks!
--b.
SUNRPC: Change TCP socket space reservation
The current server rpc tcp code attempts to predict how much writeable
socket space will be available to a given RPC call before accepting it
for processing. On a 40GigE network, we've found this throttles
individual clients long before the network or disk is saturated. The
server may handle more clients easily, but the bandwidth of individual
clients is still artificially limited.
Instead of trying (and failing) to predict how much writeable socket space
will be available to the RPC call, just fall back to the simple model of
deferring processing until the socket is uncongested.
This may increase the risk of fast clients starving slower clients; in
such cases, the previous patch allows setting a hard per-connection
limit.
next prev parent reply other threads:[~2016-07-06 20:17 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-06-24 14:55 [PATCH 01/10] SUNRPC: Fix tracepoint storage issues with svc_recv and svc_rqst_status Trond Myklebust
2016-06-24 14:55 ` [PATCH 02/10] SUNRPC: Don't allocate a full sockaddr_storage for tracing Trond Myklebust
2016-06-24 14:55 ` [PATCH 03/10] SUNRPC: Add a tracepoint for server socket out-of-space conditions Trond Myklebust
2016-06-24 14:55 ` [PATCH 04/10] SUNRPC: Add tracepoints for dropped and deferred requests Trond Myklebust
2016-06-24 14:55 ` [PATCH 05/10] SUNRPC: lock the socket while detaching it Trond Myklebust
2016-06-24 14:55 ` [PATCH 06/10] SUNRPC: Call the default socket callbacks instead of open coding Trond Myklebust
2016-06-24 14:55 ` [PATCH 07/10] SUNRPC: Micro optimisation for svc_data_ready Trond Myklebust
2016-06-24 14:55 ` [PATCH 08/10] SUNRPC: Add a server side per-connection limit Trond Myklebust
2016-06-24 14:55 ` [PATCH 09/10] SUNRPC: Change TCP socket space reservation Trond Myklebust
2016-06-24 14:55 ` [PATCH 10/10] SUNRPC: Remove unused callback xpo_adjust_wspace() Trond Myklebust
2016-06-24 21:18 ` [PATCH 09/10] SUNRPC: Change TCP socket space reservation J. Bruce Fields
2016-06-24 21:31 ` Trond Myklebust
2016-07-06 20:17 ` Fields Bruce [this message]
2016-07-06 19:53 ` J. Bruce Fields
2016-07-06 20:11 ` Trond Myklebust
2016-07-06 20:30 ` Fields Bruce
2016-06-24 21:15 ` [PATCH 08/10] SUNRPC: Add a server side per-connection limit J. Bruce Fields
2016-06-24 21:24 ` Trond Myklebust
2016-07-06 16:16 ` J. Bruce Fields
2016-06-24 21:06 ` [PATCH 05/10] SUNRPC: lock the socket while detaching it J. Bruce Fields
2016-06-24 21:21 ` Trond Myklebust
2016-06-24 18:28 ` [PATCH 01/10] SUNRPC: Fix tracepoint storage issues with svc_recv and svc_rqst_status J. Bruce Fields
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20160706201748.GG18856@fieldses.org \
--to=bfields@fieldses.org \
--cc=linux-nfs@vger.kernel.org \
--cc=trondmy@primarydata.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).