DPDK patches and discussions
 help / color / mirror / Atom feed
From: Stephen Hemminger <stephen@networkplumber.org>
To: Ruitao Gong <ruitao.gong@mavenir.com>
Cc: "vpp-dev@lists.fd.io" <vpp-dev@lists.fd.io>,
	" dev@dpdk.org" <dev@dpdk.org>,
	"gong790802715@163.com" <gong790802715@163.com>
Subject: Re: VPP main thread blocked by DPDK APIs
Date: Wed, 22 Dec 2021 09:02:17 -0800	[thread overview]
Message-ID: <20211222090217.2ce6c420@hermes.local> (raw)
In-Reply-To: <DM6PR11MB4076C5F6C995E1190E3DFE58E47D9@DM6PR11MB4076.namprd11.prod.outlook.com>

On Wed, 22 Dec 2021 08:25:49 +0000
Ruitao Gong <ruitao.gong@mavenir.com> wrote:

> hi DEVers,
> 
> Send this mail for some help about VPP and DPDK.
> rte_eth_stats_get (xd->port_id, &xd->stats)
> rte_eth_xstats_get (xd->port_id, xd->xstats, len)
>     APIs in dpdk_update_counters() node "dpdk_process" will cost about 140ms each physical interface every 10s (DPDK_STATS_POLL_INTERVAL).
>     main_thread will blocked N*140 ms when N physical interface in up state.
>     if govpp send heartbeat message to VPP, main_thread will have chance delay to read msg and reply, when main_thread blocked by doing update_counters.
>     link between GoVPP and VPP will turn down when delay time more than 250ms. It is easy to reproduce the scenario.
>     is there any plan to fix it or some suggestion to avoid it? Many thanks!
> 
> env details:
>     GoVPP:v0.3.5
>     DPDK version: 20.05
>     VPP version:20.09
>     centos7 + Intel X710
> expansion-rom-version:
> bus-info: 0000:19:00.0
> supports-statistics: yes
> supports-test: yes
> supports-eeprom-access: yes
> supports-register-dump: yes
> supports-priv-flags: yes
> 
>     startup.conf dpdk uio-driver vfio_pci
>     driver: i40evf version: 3.2.2-k
> 
> B&R,
> Ruitao

Polling for statistics may be expensive based on the hardware, it requires the device driver
to query the hardware for each queue. These requests may involve setting a flag and polling
for completion of the request.

We had similar issues, with bnxt; ended up doing small changes to driver to avoid
polling for queue information for unused queues and that helped. But getting statistics
can be still be too expensive.


> ________________________________
> This e-mail message may contain confidential or proprietary information of Mavenir Systems, Inc. or its affiliates and is intended solely for the use of the intended recipient(s). If you are not the intended recipient of this message, you are hereby notified that any review, use or distribution of this information is absolutely prohibited and we request that you delete all copies in your control and contact us by e-mailing to security@mavenir.com. This message contains the views of its author and may not necessarily reflect the views of Mavenir Systems, Inc. or its affiliates, who employ systems to monitor email messages, but make no representation that such messages are authorized, secure, uncompromised, or free from computer viruses, malware, or other defects. Thank You

Please lose the header or suppress it. Either your main should never be on mailing list because
it violates that clause, or the clause is irrelevant.

      reply	other threads:[~2021-12-22 17:02 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-22  8:25 Ruitao Gong
2021-12-22 17:02 ` Stephen Hemminger [this message]

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=20211222090217.2ce6c420@hermes.local \
    --to=stephen@networkplumber.org \
    --cc=dev@dpdk.org \
    --cc=gong790802715@163.com \
    --cc=ruitao.gong@mavenir.com \
    --cc=vpp-dev@lists.fd.io \
    /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).