From: Mohamed Mohamed <msherif4@yahoo.com>
To: "users@dpdk.org" <users@dpdk.org>
Subject: [dpdk-users] Iv40evf performance degradation when upgrade DPDK to 19-11
Date: Thu, 22 Oct 2020 13:12:11 +0000 (UTC) [thread overview]
Message-ID: <789541096.1689721.1603372331229@mail.yahoo.com> (raw)
In-Reply-To: <789541096.1689721.1603372331229.ref@mail.yahoo.com>
Hi All:
we are in the process of upgrading DPDK to 19-11 while testing i40evf driver with multiple queues we are seeing performance drop significantly the driver log seems mostly clean 10-21-2020 15:53:26.474906 i40evf_handle_aq_msg(): Failed to read msg from AdminQ,ret: -5710-21-2020 15:53:26.635553 i40evf_check_api_version(): Peer is Linux PF host10-21-2020 15:53:26.645551 EAL: PCI device 0000:00:0d.0 on NUMA socket -110-21-2020 15:53:26.645554 EAL: Invalid NUMA socket, default to 010-21-2020 15:53:26.645556 EAL: probe driver: 8086:154c net_i40e_vf10-21-2020 15:53:26.695593 i40evf_handle_aq_msg(): Failed to read msg from AdminQ,ret: -5710-21-2020 15:53:26.858821 i40evf_check_api_version(): Peer is Linux PF host10-21-2020 15:53:26.868819 EAL: PCI device 0000:00:0e.0 on NUMA socket -110-21-2020 15:53:26.868821 EAL: Invalid NUMA socket, default to 010-21-2020 15:53:26.868823 EAL: probe driver: 8086:154c net_i40e_vf10-21-2020 15:53:26.918849 i40evf_handle_aq_msg(): Failed to read msg from AdminQ,ret: -5710-21-2020 15:53:27.81824 i40evf_check_api_version(): Peer is Linux PF host10-21-2020 15:53:27.124821 set_rxtx_funcs(): virtio: using standard Tx path on port 010-21-2020 15:53:27.124853 set_rxtx_funcs(): virtio: using mergeable buffer Rx path on port 010-21-2020 15:53:27.125810 set_rxtx_funcs(): virtio: using standard Tx path on port 110-21-2020 15:53:27.125822 set_rxtx_funcs(): virtio: using mergeable buffer Rx path on port 110-21-2020 15:53:27.126606 set_rxtx_funcs(): virtio: using standard Tx path on port 210-21-2020 15:53:27.126615 set_rxtx_funcs(): virtio: using mergeable buffer Rx path on port 210-21-2020 15:53:27.141849 i40evf_handle_aq_msg(): Failed to read msg from AdminQ,ret: -57
from DPDK stats I see rx_dropped_packets : 7423883
rx_missed_errors : 7505403
in one run I have seen more errors in the log but that happened for me once not consistent 10-20-2020 13:49:59.217581 i40evf_stop_queues(): Fail to stop queue 010-20-2020 13:49:59.217582 i40evf_execute_vf_cmd(): fail to send cmd 910-20-2020 13:49:59.217584 i40evf_switch_queue(): fail to switch RX 1 off10-20-2020 13:49:59.217585 i40evf_dev_rx_queue_stop(): Failed to switch RX queue 1 off10-20-2020 13:49:59.217587 i40evf_stop_queues(): Fail to stop queue 110-20-2020 13:49:59.217588 i40evf_execute_vf_cmd(): fail to send cmd 910-20-2020 13:49:59.217590 i40evf_switch_queue(): fail to switch RX 2 off10-20-2020 13:49:59.217591 i40evf_dev_rx_queue_stop(): Failed to switch RX queue 2 off
I can't build any theory based on the above one time instance
Can you pls provide some guidance how we can debug this further and how ?
Thanks,Mohamed
parent reply other threads:[~2020-10-22 13:12 UTC|newest]
Thread overview: expand[flat|nested] mbox.gz Atom feed
[parent not found: <789541096.1689721.1603372331229.ref@mail.yahoo.com>]
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=789541096.1689721.1603372331229@mail.yahoo.com \
--to=msherif4@yahoo.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).