Skip to content

[doc] Add README for a basic PFC watchdog test#2336

Merged
lguohan merged 2 commits intosonic-net:masterfrom
baiwei0427:pfcwd-docs
Oct 15, 2020
Merged

[doc] Add README for a basic PFC watchdog test#2336
lguohan merged 2 commits intosonic-net:masterfrom
baiwei0427:pfcwd-docs

Conversation

@baiwei0427
Copy link
Copy Markdown
Contributor

Description of PR

Signed-off-by: Wei Bai webai@microsoft.com

Summary: README file for a basic PFC watchdog test.
Fixes # (issue)

Type of change

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

Approach

What is the motivation for this PR?

How did you do it?

How did you verify/test it?

Any platform specific information?

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

Documentation


- When $T_{storm}$ is larger than $T_{detect}$ + $T_{poll}$:
- PFC watchdog is triggered on the corresponding lossless priority of et2.
- All the packets of traffic 1 are dropped.
Copy link
Copy Markdown
Contributor

@neethajohn neethajohn Oct 14, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this under the assumption that the PFC pause storm will entirely pause the queue? if not, some packets will still make it through till the storm is detected

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this under the assumption that the PFC pause storm will entirely pause the queue? if not, some packets will still make it through till the storm is detected

Yes. See description of PFC storm.

Copy link
Copy Markdown
Contributor

@SuvenduMozumdar SuvenduMozumdar Oct 14, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The Traffic 1 is started at T_restore/2. I believe this is smaller than T_detect+T_poll, in which case some initial traffic will be passed through. Which does not match with the expectation that Traffic 1 should be 100% dropped.

Traffic 1 should start at (T_detect+T_poll) and run for the duration (T_storm - T_detect - T_poll + T_restore/2). This will make sure 100% traffic drop for Traffic 1.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The Traffic 1 is started at T_restore/2. I believe this is smaller than T_detect+T_poll, in which case some initial traffic will be passed through. Which does not match with the expectation that Traffic 1 should be 100% dropped.

Traffic 1 should start at (T_detect+T_poll) and run for the duration (T_storm - T_detect - T_poll + T_restore/2). This will make sure 100% traffic drop for Traffic 1.

This is wrong. Though traffic 1 starts before the triggering of PFC watchdog, it will be blocked by the PFC storm. Once PFC watchdog is triggered, all the packets queued in the switch will be dropped.

@lguohan lguohan merged commit 0bc30b0 into sonic-net:master Oct 15, 2020
@baiwei0427 baiwei0427 deleted the pfcwd-docs branch February 4, 2021 09:03
kazinator-arista pushed a commit to kazinator-arista/sonic-mgmt that referenced this pull request Mar 4, 2026
Include following commits:
```
839625d Replace cmp in acl_loader with operator.eq (sonic-net#2328)
af245ad [route_check]: Ignore standalone tunnel routes (sonic-net#2332)
c9aa65c Revert "[DHCPv6] [202012] Update the dhcpv6_relay config/show cli (sonic-net#2271)" (sonic-net#2336)
6b9cdc9 Made changes to only allow a valid IPv4 or Ipv6 string for dhcp cli (sonic-net#2296)
```
kazinator-arista pushed a commit to kazinator-arista/sonic-mgmt that referenced this pull request Mar 4, 2026
7272bf37 Revert "Revert "[DHCPv6] [202012] Update the dhcpv6_relay config/show cli (sonic-net#2271)" (sonic-net#2336)" (sonic-net#2459)
Signed-off-by: Vivek Reddy <vkarri@nvidia.com>
kazinator-arista pushed a commit to kazinator-arista/sonic-mgmt that referenced this pull request Mar 4, 2026
…et#11254)

swss:
* ad2d0ad 2022-06-24 | [PFC_WD] Avoid applying ZeroBuffer Profiles to ingress PG when a PFC storm is detected (sonic-net#2304) (HEAD -> 202205) [Vivek R]
* ef75554 2022-06-25 | [swssconfig] Optimize performance of swssconfig (sonic-net#2336) [Junchao-Mellanox]
* d9e9ba8 2022-06-24 | [fdborch] fix heap-use-after-free in clearFdbEntry() (sonic-net#2353) [Yakiv Huryk]
* 585a69b 2022-06-24 | Create ACL table fails due to incorrect check for supported ACL actions sonic-net#11235 (sonic-net#2351) [Ravindranath C K]
* 0d19560 2022-06-24 | [macsec] Refactor the logic of macsec name map (sonic-net#2348) [Junhua Zhai]
* 111dfc2 2022-06-23 | [macsec] Update macsec flex counter (sonic-net#2338) (HEAD -> 202205, github/202205) [Junhua Zhai]

swss-common:
* 0213d55 2022-06-23 | [portcounter] Check if counter ID exists before arithmetic operation  (sonic-net#632) (HEAD -> 202205, github/202205) [Junhua Zhai]
* c21c47e 2022-06-14 | [counter] Add counter table (sonic-net#622) [Junhua Zhai]

utilities:
* 430cd65 2022-06-23 | [202205] [generate dump] Move the Core/Log collection to the End of process Execution and removed default timeout  (sonic-net#2230) (github/202205) [Vivek R]

linkmgrd:
* 59334be 2022-06-24 | Remove exception throwing when initializing missing loopback interface (sonic-net#90) (HEAD -> 202205) [Jing Zhang]

Signed-off-by: Ying Xie <ying.xie@microsoft.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants