Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Content-Type: text/plain;
        charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 12.2 \(3445.102.3\))
Subject: Re: [PATCH] mlx4_ib: Increase the timeout for CM cache
From:   =?utf-8?Q?H=C3=A5kon_Bugge?= <haakon.bugge@oracle.com>
In-Reply-To: <20190131170951.178676-1-haakon.bugge@oracle.com>
Date:   Fri, 1 Feb 2019 16:18:34 +0100
Cc:     OFED mailing list <linux-rdma@vger.kernel.org>,
        linux-kernel@vger.kernel.org
Content-Transfer-Encoding: quoted-printable
Message-Id: <9BDCB055-BBB7-497F-BA28-F092051E4B22@oracle.com>
References: <20190131170951.178676-1-haakon.bugge@oracle.com>
To:     Doug Ledford <dledford@redhat.com>,
        Sean Hefty <sean.hefty@intel.com>,
        Hal Rosenstock <hal@dev.mellanox.co.il>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Sorry,

I posted this as if it a net patch. It isn't, hence resend with correct =
recipients.


Thxs, H=C3=A5kon


> On 31 Jan 2019, at 18:09, H=C3=A5kon Bugge <haakon.bugge@oracle.com> =
wrote:
>=20
> Using CX-3 virtual functions, either from a bare-metal machine or
> pass-through from a VM, MAD packets are proxied through the PF driver.
>=20
> Since the VMs have separate name spaces for MAD Transaction Ids
> (TIDs), the PF driver has to re-map the TIDs and keep the book keeping
> in a cache.
>=20
> Following the RDMA CM protocol, it is clear when an entry has to
> evicted form the cache. But life is not perfect, remote peers may die
> or be rebooted. Hence, it's a timeout to wipe out a cache entry, when
> the PF driver assumes the remote peer has gone.
>=20
> We have experienced excessive amount of DREQ retries during fail-over
> testing, when running with eight VMs per database server.
>=20
> The problem has been reproduced in a bare-metal system using one VM
> per physical node. In this environment, running 256 processes in each
> VM, each process uses RDMA CM to create an RC QP between himself and
> all (256) remote processes. All in all 16K QPs.
>=20
> When tearing down these 16K QPs, excessive DREQ retries (and
> duplicates) are observed. With some cat/paste/awk wizardry on the
> infiniband_cm sysfs, we observe:
>=20
>      dreq:       5007
> cm_rx_msgs:
>      drep:       3838
>      dreq:      13018
>       rep:       8128
>       req:       8256
>       rtu:       8256
> cm_tx_msgs:
>      drep:       8011
>      dreq:      68856
>       rep:       8256
>       req:       8128
>       rtu:       8128
> cm_tx_retries:
>      dreq:      60483
>=20
> Note that the active/passive side is distributed.
>=20
> Enabling pr_debug in cm.c gives tons of:
>=20
> [171778.814239] <mlx4_ib> mlx4_ib_multiplex_cm_handler: id{slave:
> 1,sl_cm_id: 0xd393089f} is NULL!
>=20
> By increasing the CM_CLEANUP_CACHE_TIMEOUT from 5 to 30 seconds, the
> tear-down phase of the application is reduced from 113 to 67
> seconds. Retries/duplicates are also significantly reduced:
>=20
> cm_rx_duplicates:
>      dreq:       7726
> []
> cm_tx_retries:
>      drep:          1
>      dreq:       7779
>=20
> Increasing the timeout further didn't help, as these duplicates and
> retries stem from a too short CMA timeout, which was 20 (~4 seconds)
> on the systems. By increasing the CMA timeout to 22 (~17 seconds), the
> numbers fell down to about one hundred for both of them.
>=20
> Adjustment of the CMA timeout is _not_ part of this commit.
>=20
> Signed-off-by: H=C3=A5kon Bugge <haakon.bugge@oracle.com>
> ---
> drivers/infiniband/hw/mlx4/cm.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>=20
> diff --git a/drivers/infiniband/hw/mlx4/cm.c =
b/drivers/infiniband/hw/mlx4/cm.c
> index fedaf8260105..8c79a480f2b7 100644
> --- a/drivers/infiniband/hw/mlx4/cm.c
> +++ b/drivers/infiniband/hw/mlx4/cm.c
> @@ -39,7 +39,7 @@
>=20
> #include "mlx4_ib.h"
>=20
> -#define CM_CLEANUP_CACHE_TIMEOUT  (5 * HZ)
> +#define CM_CLEANUP_CACHE_TIMEOUT  (30 * HZ)
>=20
> struct id_map_entry {
> 	struct rb_node node;
> --=20
> 2.20.1
>=20