Content-Type: text/plain;
        charset=us-ascii
Mime-Version: 1.0 (Mac OS X Mail 11.5 \(3445.9.1\))
Subject: Re: mount option timeo - in deciseconds, or seconds?
From:   Chuck Lever <chuck.lever@oracle.com>
In-Reply-To: <20181215025034.j5zyfytvxqt35woz@mayhem.atnf.CSIRO.AU>
Date:   Mon, 17 Dec 2018 10:11:34 -0500
Cc:     Linux NFS Mailing List <linux-nfs@vger.kernel.org>
Content-Transfer-Encoding: quoted-printable
Message-Id: <8EE078EE-CE10-4316-B447-F005B2034D69@oracle.com>
References: <20181215025034.j5zyfytvxqt35woz@mayhem.atnf.CSIRO.AU>
To:     Vincent McIntyre <vincent.mcintyre@csiro.au>
Sender: linux-nfs-owner@vger.kernel.org
Precedence: bulk

Hi Vincent-


> On Dec 14, 2018, at 9:50 PM, Vincent McIntyre =
<vincent.mcintyre@csiro.au> wrote:
>=20
> Hi,
>=20
> I've been trying to figure out why the nfs.man manpage states
>=20
>  .BI timeo=3D n
>  The time in deciseconds (tenths of a second) the NFS client waits for =
a
>  response before it retries an NFS request.
>=20
> Can someone who actually understands NFS enlighten me?
> Here's where I'm up to.
>=20
> I looked in the upstream git.
> The 'deciseconds' has been in there since the start of git history
> (2007, commit id 16db99b56a532bf56fa27618a6ef30763cd9006f).
> But is it (still) correct?
>=20
> Unfortunately I can't see a direct use of the 'timeo' field
> of the NFS data structure so the answer is not easily found.

The "timeo" mount option is not used by the mount.nfs command.
It is passed to the kernel, which treats it just as the man
page says.

The code you excerpt below implements the "retry" mount option.

.BI retry=3D n
The number of minutes that the
.BR mount (8)
command retries an NFS mount operation
in the foreground or background before giving up.
If this option is not specified, the default value for foreground mounts
is 2 minutes, and the default value for background mounts is 10000 =
minutes
(80 minutes shy of one week).
If a value of zero is specified, the
.BR mount (8)
command exits immediately after the first failure.
.IP
Note that this only affects how many retries are made and doesn't
affect the delay caused by each retry.  For UDP each retry takes the
time determined by the
.BR timeo
and
.BR retrans
options, which by default will be about 7 seconds.  For TCP the
default is 3 minutes, but system TCP connection timeouts will=20
sometimes limit the timeout of each retransmission to around 2 minutes.


> In utils/mount/nfsmount.c mountnfs() we have
>=20
> timeout =3D time(NULL) + 60 * retry;
>=20
> This sets the upper bound for how long to wait for a mount to succeed.
> The units are seconds, because retry is given in minutes.
>=20
> Also we see
>=20
> prevt =3D 0;
> t =3D 30;
> val =3D 1;
>=20
> these are the control variables for the loop
> The main check that prolongs the waiting time is
>=20
>            if (t - prevt < 30)
>                sleep(30);
>             ....if not mounted yet...
>            prevt =3D t;
>=20
> The decision to time out is made like this
>            t =3D time(NULL);
>            if (t >=3D timeout) {
>                rpc_mount_errors(*nfs_server.hostname, 0, bg);
>                goto fail;
>            }
>=20
> so it seems clear that t should also have the units of seconds.
> That also matches up with the declaration of t & timeout as time_t.
>=20
> But I still don't see how data.timo gets used.
> utils/mount/network.h defines this
>=20
>  static const struct timeval TIMEOUT =3D { 20, 0 };
> and
>  #define MOUNT_TIMEOUT   (30)
>=20
> which would seem to be in units of seconds,
> since eg. connect_to() uses seconds
> which reaches it via the calling path:
>=20
> 1.
> CLIENT *mnt_openclnt(clnt_addr_t *mnt_server, int *msock)
> {
> ...
>    *msock =3D get_socket(mnt_saddr, mnt_pmap->pm_prot, MOUNT_TIMEOUT,
>                TRUE, FALSE);
>=20
> 2.
> static int get_socket(struct sockaddr_in *saddr, unsigned int p_prot,
>            unsigned int timeout, int resvp, int conn)
> {
> ...
>        cc =3D connect_to(so, SAFE_SOCKADDR(saddr), namelen,
>                timeout);
> 3.
> /*
> * Attempt to connect a socket, but time out after "timeout" seconds.
> *
> * On error return, caller closes the socket.
> */
> static int connect_to(int fd, struct sockaddr *addr,
>            socklen_t addrlen, int timeout)
> {
>=20
>=20
> Thanks
> Vince

--
Chuck Lever