Skip to content

[action] [PR:13119] Fix qos/test_qos_dscp_mapping.py#13260

Merged
mssonicbld merged 1 commit intosonic-net:202311from
mssonicbld:cherry/202311/13119
Jun 13, 2024
Merged

[action] [PR:13119] Fix qos/test_qos_dscp_mapping.py#13260
mssonicbld merged 1 commit intosonic-net:202311from
mssonicbld:cherry/202311/13119

Conversation

@mssonicbld
Copy link
Collaborator

Description of PR

Summary: Fix qos/test_qos_dscp_mapping.py failures
Fixes # aristanetworks/sonic-qual.msft#132

Type of change

  • Bug fix
  • Testbed and Framework(new/improvement)
  • Test case(new/improvement)

Back port request

  • 201911
  • 202012
  • 202205
  • 202305
  • 202311

Approach

What is the motivation for this PR?

This test is mainly failing due to:

  • Sending 10K packets at once on a ptf interface is sometimes causing nnpy socket to be non-functional for a moment and connection is getting timedout (logic to timeout is introduced by Add a send timeout to nnpy sockets. p4lang/ptf#183).
  • Sometimes egress_queue_count on queue 7 is way greater than no of packets being sent and the queue_count is not within the TOLERANCE, this could be due to protocol packets.
  • Test is not enhanced to be run on dualtor topologies (packets are going to unexpected ToR).
  • Fixtures "upstream_links" and "downstream_links" are always referring to upper ToR, thus return values of these fixtures is not really considering the ToR type and causing failures.

How did you do it?

Fixes involve adding right fixtures to the test methods (to consider dualtor topologies) and minor changes to address above issues.

How did you verify/test it?

With the fix test is passing well (earlier it was failing most of the times with one or other reason). Verified the fix on Arista-7260CX3-C64 202311 branch.

Any platform specific information?

Supported testbed topology if it's a new test case?

Documentation

* Fix qos/test_qos_dscp_mapping.py

This test is mainly failing due to:
   - Sending 10K packets at once on a ptf interface is causing nnpy
     socket to be non-functional for a moment sometimes and connection
     is getting timedout.
   - Sometimes egress_queue_count on queue 7 is way greater than no of
     packets being sent and the queue_count is not within the TOLERANCE,
     this could be due to protocol packets.
   - Test is not enhanced to be run on dualtor topologies (packets are
     going to unexpected ToR).
   - Fixtures "upstream_links" and "downstream_links" are always
     referring to upper ToR, thus return values of these fixtures is not
     really considering the ToR type and causing failures.

* Fix pre-commit check failures

pre-commit check is being failed with
"toggle_all_simulator_ports_to_rand_selected_tor" imported but unused
error. Added "# noqa" to suppress this failure as this import is being
used as fixture argument (similar to other tests).
@mssonicbld
Copy link
Collaborator Author

Original PR: #13119

@mssonicbld mssonicbld merged commit a5f6d94 into sonic-net:202311 Jun 13, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants