DPDK usage discussions
 help / color / mirror / Atom feed
From: "Pandey, Utkarsh" <Utkarsh.Pandey@hughes.com>
To: "users@dpdk.org" <users@dpdk.org>
Subject: Problem creating a bond in active backup mode on Virtual Functions on Intel X710 network controller
Date: Wed, 29 Jun 2022 02:38:59 +0000	[thread overview]
Message-ID: <DM5PR1101MB2169E3934D26BC0F4EBF8BC2FABB9@DM5PR1101MB2169.namprd11.prod.outlook.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 1717 bytes --]

Hello,
We have two X710 network interfaces (physical functions) PF1 and PF2 and creating one virtual functions (VF) on each of them (say VF1 and VF2 respectively).
We are creating a bond between the two VFs (VF1 and VF2 belonging to PF1 and PF2 respectively), in active-backup mode using EAL arguments and assigning a bond mac address.
Initially VF1 is the primary slave and VF2 is the backup.
We have been successful in using this bond interface to send and receive packets.

However, when we test for redundancy, we are running into a problem described as follows:
1. Initially the bond interface is able to send network packets using the primary slave.
2. The primary slave (VF1) is brought down by bringing down the physical functions (PF1) corresponding to the primary slave.
3. The backup VF (VF2) becomes the primary slave and the bond interface continues to transmit packets.
4. Some time later, the original primary slave (VF1) is re-enabled and the current slave (VF2) is brought down.
5. The primary slave for the bond interface changes back to the initial interface (VF1). However, at this point, the packet transmission stops.

Experiment Details:
1. DPDK 21.05 is used.
2. The physical functions are bound to i40e driver and the virtual functions are bound to the vfio-pci driver.
3. The VFs were not assigned a MAC address through the PF manually.
4. Trust mode was set to "On" for both the VFs.

A similar experiment performed on the Physical Function does not show this issue and the primary slave switches back and forth as the interfaces are disabled and re-enabled.

Is it possible to create a bond on Virtual Functions in the active-backup mode?

Thanks,
Utkarsh Pandey.

[-- Attachment #2: Type: text/html, Size: 3611 bytes --]

                 reply	other threads:[~2022-06-29  2:39 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

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=DM5PR1101MB2169E3934D26BC0F4EBF8BC2FABB9@DM5PR1101MB2169.namprd11.prod.outlook.com \
    --to=utkarsh.pandey@hughes.com \
    --cc=users@dpdk.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 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).