All of lore.kernel.org
 help / color / mirror / Atom feed
From: Kay Sievers <kay.sievers@vrfy.org>
To: linux-hotplug@vger.kernel.org
Subject: Re: [PATCH][RFC] rules: add persistent vf mac rules generator for
Date: Thu, 18 Aug 2011 11:13:22 +0000	[thread overview]
Message-ID: <CAPXgP13yYeJfgOjSUA637C-hgCp6Bs5yzV+tdfXdRDUv+qxAhQ@mail.gmail.com> (raw)
In-Reply-To: <1313663068-26397-1-git-send-email-lidongyang@novell.com>

On Thu, Aug 18, 2011 at 12:24, Li Dongyang <lidongyang@novell.com> wrote:
> This is the rule generator for SR-IOV device intel 82576,
> because the mac address of each Virtual Function interfaces are generated
> randomly each time the igb driver loaded, this is a way to make the
> mac of the VFs consistent cross reboots.
> What we do is when we see the VFs up first time, we record each mac
> and generate a rule, writing each mac into the rule and set it via ip utility
> on the next boot.
>
> and when we generate the rule, we block and wait on the files under /sys
> cause sometimes the udev event is triggerd before the VFs are up completely,
> and we might see no mac of the VFs if we do not block. However, during
> installation of a system like SLES, the rule generator will stall
> and wait for the file, so I think some suggestions are needed, Thanks

We are about to kill the entire rule_generator in the next months. It
has too man problems, and does not really solve any real world
problem:
The majority of systems does not rely on predictable names, and does
not need any automatic rules mangling from the hotplug path.

The systems who rely on predictable names need manual configuration
anyway. So we will just require explicit configuration for predictable
names instead of automatically generating system config from the
hotplug path.

The rule_generator cause too many problems to continue it. The
intersection of systems who need it and the systems which do not have
manual config anyway is almost zero, so it makes not much sense to
continue that road.

The current plan is to have a generic format for network configuration
and create udev rules and systemd service files on-the-fly at bootup,
before udev is even started. The generated udev rules will trigger
when devices show up in the hotplug path and rename the interfaces.
Systemd will start a service for the every configured interface which
will apply the needed IP config or start things like the dhcp client.
Unconfigured interfaces without a specific matching configuration file
will just keep the kernel's name, there will be no try to create any
persistent records for it.

This all is expected to happen during the next months, and the code to
do this will likely be merged into systemd and not udev. The udev rule
generator will be deleted in the longer run. We tried, but we need to
admit today, that it isn't the solution we are looking for, to solve
issues regarding predictable network interface names.

Thanks,
Kay

  reply	other threads:[~2011-08-18 11:13 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-08-18 10:24 [PATCH][RFC] rules: add persistent vf mac rules generator for SR-IOV device intel 82576 Li Dongyang
2011-08-18 11:13 ` Kay Sievers [this message]
2011-08-18 22:58 ` [PATCH][RFC] rules: add persistent vf mac rules generator for Marco d'Itri

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=CAPXgP13yYeJfgOjSUA637C-hgCp6Bs5yzV+tdfXdRDUv+qxAhQ@mail.gmail.com \
    --to=kay.sievers@vrfy.org \
    --cc=linux-hotplug@vger.kernel.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.