Update Arista7060X6-PE 256x200G mmu configuration. #19570
Conversation
|
/azpw ms_conflict |
| "egress_lossless_profile": { | ||
| "pool": "egress_lossless_pool", | ||
| "size": "0", | ||
| "static_th": "125995684" |
There was a problem hiding this comment.
why is this larger than the egress_lossless pool size?
There was a problem hiding this comment.
I checked the configuration and also the update today. This setting is indeed larger than egress pool. I have replied the CSP case to confirm the reason, hopefully will get answers soon.
|
|
||
| {%- macro generate_buffer_pool_and_profiles() %} | ||
| "BUFFER_POOL": { | ||
| "ingress_lossless_pool": { |
There was a problem hiding this comment.
instead of lossless pool, should we say this is ingress_pool?
There was a problem hiding this comment.
currently, the qos tests are using these names for grabbing the configurations, so changing the name will cause them to break.
| "mode": "dynamic", | ||
| "xoff": "29520896" | ||
| }, | ||
| "egress_lossless_pool": { |
There was a problem hiding this comment.
should we just say this is egress_pool?
There was a problem hiding this comment.
i do not know if other repo (swss/sonic-utilitie) make any assumption on the name, we need to check that.
There was a problem hiding this comment.
yes, things like qos tests are currently assuming the name while grabing the configuration, changing the name will cause these tests to break at this moment...
|
/azpw ms_conflict |
| @@ -8,7 +8,7 @@ | |||
| "size": "165660324", | |||
There was a problem hiding this comment.
I believe these values should now be reverted back to reflect the production bcm config, e.g. 121964196 here
There was a problem hiding this comment.
not all numbers, these values are the latest recommended ones.
There was a problem hiding this comment.
b62e111#diff-38141bc38cf22f25e51d1afa26023349fcb218e27d0c46e8ada19c6c9081111cL2251-L2252
I'm not sure how 165660324 was derived in the second commit, but I believe this change to the SHARED_LIMIT_CELLS needs to be reflected? (240087 * 254 * 2 = 121964196)
There was a problem hiding this comment.
I think we are aligned from calculation perspective, however if you check the sai api values that shared by broadcom. the value is indeed this....
|
/azpw ms_conflict |
…et#19570) Why I did it This PR updates the MMU related configurations on Arista 7060X6-PE device with 256x200G breakout. Work item tracking Microsoft ADO (number only): 28707303 How I did it This PR updates 3 things: - Updated bcm file for optimal MMU settings. - Updated buffer defaults to accommodate the TH5 architecture with 1 ingress pool + 1 egress pool and updated to optimal value. - Updated PG lookups for buffer setups. How to verify it Tested with sonic-mgmt tests with xoff/xon tests with updated QoS parameter: https://github.com/sonic-net/sonic-mgmt/pull/13656/files Local ixia test is passing in lab. Both verified using 202311 branch for backporting.
|
Cherry-pick PR to 202311: #19604 |
Why I did it This PR updates the MMU related configurations on Arista 7060X6-PE device with 256x200G breakout. Work item tracking Microsoft ADO (number only): 28707303 How I did it This PR updates 3 things: - Updated bcm file for optimal MMU settings. - Updated buffer defaults to accommodate the TH5 architecture with 1 ingress pool + 1 egress pool and updated to optimal value. - Updated PG lookups for buffer setups. How to verify it Tested with sonic-mgmt tests with xoff/xon tests with updated QoS parameter: https://github.com/sonic-net/sonic-mgmt/pull/13656/files Local ixia test is passing in lab. Both verified using 202311 branch for backporting.
mssonicbld
added
Included in 202311 Branch
and removed
Created PR to 202311 Branch
labels
…et#19570) Why I did it This PR updates the MMU related configurations on Arista 7060X6-PE device with 256x200G breakout. Work item tracking Microsoft ADO (number only): 28707303 How I did it This PR updates 3 things: - Updated bcm file for optimal MMU settings. - Updated buffer defaults to accommodate the TH5 architecture with 1 ingress pool + 1 egress pool and updated to optimal value. - Updated PG lookups for buffer setups. How to verify it Tested with sonic-mgmt tests with xoff/xon tests with updated QoS parameter: https://github.com/sonic-net/sonic-mgmt/pull/13656/files Local ixia test is passing in lab. Both verified using 202311 branch for backporting.
…et#19570) Why I did it This PR updates the MMU related configurations on Arista 7060X6-PE device with 256x200G breakout. Work item tracking Microsoft ADO (number only): 28707303 How I did it This PR updates 3 things: - Updated bcm file for optimal MMU settings. - Updated buffer defaults to accommodate the TH5 architecture with 1 ingress pool + 1 egress pool and updated to optimal value. - Updated PG lookups for buffer setups. How to verify it Tested with sonic-mgmt tests with xoff/xon tests with updated QoS parameter: https://github.com/sonic-net/sonic-mgmt/pull/13656/files Local ixia test is passing in lab. Both verified using 202311 branch for backporting.
|
@r12f Please file another PR for 202405 branch to address cherry-pick conflict. |
Signed-off-by: Janetxxx <[email protected]>
* solve conflict in #19570 Signed-off-by: Janetxxx <[email protected]>
Why I did it
This PR updates the MMU related configurations on Arista 7060X6-PE device with 256x200G breakout.
Work item tracking
How I did it
This PR updates 3 things:
How to verify it
Both verified using 202311 branch for backporting.
Which release branch to backport (provide reason below if selected)
Tested branch (Please provide the tested image version)
Description for the changelog
Update Arista7060X6-PE 256x200G mmu configuration.
Link to config_db schema for YANG module changes
N/A
A picture of a cute animal (not mandatory but encouraged)