search

sonic-net / sonic-buildimage

Scripts which perform an installable binary image build for SONiC
Other
735 stars 1.41k forks source link

[EVPN][warm-reboot] EVPN vxlan traffic loss during warm-reboot #12370

Closed Hedgehog-Guru closed 1 year ago

Hedgehog-Guru commented 2 years ago

Description

In EVPN vxlan scenario with proper graceful restart config inside FRR, there is traffic loss during warm-reboot

Steps to reproduce the issue:

Topology: two switches connected with single port. Each switch has L2 access port to any traffic generator able to send L2 frames. Ixia in my case.

  1. Configure split config mode.
  2. Configure VTEPs on switches
  3. Configure any vlan and add member ports facing traffic generator. Create VNI for that vlan.
  4. Configure L3 addresses on interswitch link
  5. In FRR configure BGP L2VPN EVPN between switches
  6. Add FRR graceful restart options on both switches 
    bgp graceful-restart restart-time 300
bgp graceful-restart
bgp graceful-restart preserve-fw-state
  7. Start L2 traffic between switches. Make sure there are no losses
  8. Perform warm-reboot and check traffic impact

Describe the results you received:

Traffic loss

Describe the results you expected:

No any traffic impact

Output of show version:

SONiC Software Version: SONiC.202205_rc.17-12870dde8_Internal
Distribution: Debian 11.5
Kernel: 5.10.0-12-2-amd64
Build commit: 12870dde8
Build date: Tue Oct  4 09:54:37 UTC 2022
Built by: sw-r2d2-bot@r-build-sonic-ci03-242

Platform: x86_64-mlnx_msn4600c-r0
HwSKU: ACS-MSN4600C
ASIC: mellanox
ASIC Count: 1
Serial Number: MT2140X00042
Model Number: MSN4600-CS2FO_QP
Hardware Revision: A1
Uptime: 20:33:11 up 11 min,  1 user,  load average: 0.54, 0.57, 0.39
Date: Fri 07 Oct 2022 20:33:11

Output of show techsupport:

sonic_dump_qa-eth-vt03-1-4600ca1_20221007_203258.tar.gz sonic_dump_qa-eth-vt03-2-3700v_20221007_203257.tar.gz

Additional information you deem important (e.g. issue happens only occasionally):

dgsudharsan commented 2 years ago

It appears FRR doesn't support graceful restart for EVPN address family https://github.com/FRRouting/frr/issues/9749

dgsudharsan commented 2 years ago

@adyeung Can you please help to assign someone to confirm if warmboot with EVPN will not be supported in SONiC? EVPN HLD mentions that it is supported. Is it tested?

arlakshm commented 2 years ago

@yxieca to start the thread with @adyeung on EVPN warmboot issues.

adyeung commented 1 year ago

@nkelapur from BRCM will followup

nkelapur commented 1 year ago

FRR currently does not support BGP Graceful restart for EVPN address-family. Hence EVPN warm-reboot will not be hitless in Sonic until this support is available in FRR. Updated EVPN HLD with this note

Yuval-Mellanox commented 1 year ago

@nkelapur could you please point me that note? could not find it in the HLD: https://github.com/sonic-net/SONiC/blob/master/doc/vxlan/EVPN/EVPN_VXLAN_HLD.md

nkelapur commented 1 year ago

Looks like I missed creating a pull request for the proposed change. I have done it now. https://github.com/sonic-net/SONiC/pull/1227

prgeor commented 1 year ago

Not supported. Closing as hld is now updated