Bug 5811 - mlx5_core Mellanox ConnectX-5 ping has a large latency
Summary: mlx5_core Mellanox ConnectX-5 ping has a large latency
Status: NEW
Alias: None
Product: Anolis OS 8
Classification: Anolis OS
Component: kernel - anck-4.19 (show other bugs) kernel - anck-4.19
Version: 8.2
Hardware: x86_64 Linux
: P3-Medium S3-normal
Target Milestone: ---
Assignee: wuhao
QA Contact: shuming
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2023-07-10 10:59 UTC by djhwyh
Modified: 2023-09-21 15:25 UTC (History)
1 user (show)

See Also:

Attachments
ping latency (61.80 KB, image/png)
2023-07-10 10:59 UTC, djhwyh 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description djhwyh 2023-07-10 10:59:23 UTC 
Created attachment 836 [details]
ping latency

Description of problem:
不同IDC机房中的，很多服务器都有该问题，具体是同一个接入交换机下的两台龙蜥服务器互相ping，一段时间后，可能出现ping延时达几百或上千毫秒，正常是小于1毫秒的。

Version-Release number of selected component (if applicable):
操作系统名称：Anolis OS 8.2 QU1,内核：4.19.91-24.8.an8.x86_64
网卡驱动和固件信息举例：

ethtool -i eth0

driver: mlx5_core

version: 5.0-0

firmware-version: 16.32.1010 (MT_0000000248)

expansion-rom-version:

bus-info: 0000:1a:00.0

How reproducible:
同一个接入交换机下的两台龙蜥服务器互相ping，服务器里是容器业务。


Actual results:
ping延迟结果见下面附件。

Expected results:


Additional info:
Comment 1 maqiao alibaba_cloud_group 2023-07-10 11:27:29 UTC 
通过pingtrace定界分析，目前发现问题是client把包给到网卡驱动，到server的网卡驱动收到包，这个过程中有明显的延迟，怀疑是驱动或者交换机的问题
Comment 2 wuhao alibaba_cloud_group 2023-09-21 15:25:15 UTC 
换了 mellanox 官网 5.8 的驱动，并对 4.19 内核进行了适配，问题已修复