[Chassis] Added multi Asics support to everflow Ipv6 by mannytaheri · Pull Request #22237 · sonic-net/sonic-mgmt

mannytaheri · 2026-02-03T20:31:49Z

Description of PR

This is a fix for sonic-mgmt issue #21702 (All everflow/test_everflow_ipv6.py are erroring out)
This was caused by PR 21384 (Add Support for Egress V6 Testing)

Everflow issues:

Added multiAsics support - All everflow/test_everflow_ipv6.py are erroring (sonic-mgmt issue Bug: [Chassis] All everflow/test_everflow_ipv6.py are erroring ou #21702)
IP_TYPE "ANY" is not supported in SAI for DNX. Nokia created CS00012446025
IP_TYPE "IP" is installed but not hit in SAI causing traffic failure (sonic-buildimage issue #25281). Nokia created CS00012446460 - Nokia has suggested a fix to Broadcom.

Summary:
Fixes # (issue)
sonic-mgmt issue #21702

Type of change

[ x] Bug fix
Testbed and Framework(new/improvement)
New Test case
- Skipped for non-supported platforms
Test case improvement

Back port request

Approach

What is the motivation for this PR?

To Fix

How did you do it?

Added multi Asics support

How did you verify/test it?

Ran the entire everflow suites on multi Asics chassis

Any platform specific information?

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

Documentation

mssonicbld · 2026-02-03T20:32:02Z

/azp run

azure-pipelines · 2026-02-03T20:32:17Z

Azure Pipelines successfully started running 1 pipeline(s).

(cherry picked from commit 654f93c2e6adb01c785f4a18a296a67c590ea157) Signed-off-by: mtaheri <manouchehr.taheri@nokia.com>

mssonicbld · 2026-02-04T15:02:22Z

/azp run

azure-pipelines · 2026-02-04T15:02:39Z

Azure Pipelines successfully started running 1 pipeline(s).

Signed-off-by: mtaheri <manouchehr.taheri@nokia.com>

mssonicbld · 2026-02-04T16:45:32Z

/azp run

azure-pipelines · 2026-02-04T16:45:50Z

Azure Pipelines successfully started running 1 pipeline(s).

arista-nwolfe · 2026-02-04T19:32:29Z

tests/everflow/files/test_rules_ip_type_v6.json

        "{{ table_name }}|rule_998": {
            "PRIORITY": "998",
            "IP_TYPE": "IP",
            "{{ action }}": "test_session_1",
            "DST_IPV6": "2002:0225:7c6b:a982:d48b:230e:f271:0012"
        },


Do you still see everflow/test_everflow_ipv6.py::TestIngressEverflowIPv6::test_ip_type_ip failing?
It seems like IP_TYPE:IP doesn't match IPv6 packets even if the DST_IPV6 matches:

(Pdb) p mirror_packet_sent <Ether dst=9c:69:ed:ce:fa:69 src=b'fe:5c:57:27:80:00' type=IPv6 |<IPv6 tc=0 fl=0 nh=TCP hlim=64 src=2002:225:7c6b:a982:d48b:230e:f271:0 [6to4 GW: 2.37.124.107] dst=2002:225:7c6b:a982:d48b:230e:f271:12 [6to4 GW: 2.37.124.107] |<TC P sport=2020 dport=http_alt flags=A |<Raw load='tests.everflow.test_everfl' |>>>>

The testcase passes for us since I have a local fix.
IP_TYPE "IP" is installed but not hit in SAI causing traffic failure (sonic-buildimage issue #25281). Nokia created CS00012446460 - Nokia has suggested a fix to Broadcom.

I have this fix locally.

What's the reason IP_TYPE=IP should be supported by the brcm SAI but IP_TYPE=ANY shouldn't be supported by the brcm SAI?
I.E. why are we changing the test to use IP_TYPE=IPV6ANY instead of having the brcm SAI support IP_TYPE=ANY?

Both should be supported by brcm SAI. We have created CS00012446025 for ANY and CS00012446460 for IP.
Changing IP_TYPE=ANY to IP_TYPE=IPV6ANY is temporary until we have a fix from brcm.

I agree, IP_TYPE=ANY should map to IP_TYPE=IP in SAI code and we can get the fix for this from Broadcom.

Actually, reviewing https://github.com/opencomputeproject/SAI/blob/master/inc/saiacl.h#L39 IP_TYPE=ANY includes ARP request/response as well which may not be supported in DNX. Using IP_TYPE=IP should include both IPv4 and IPv6 packets and that seems to be broken now. And that's a pending SAI fix, according to @mannytaheri

IOW, the test needs to be fixed not to use ANY and SAI needs to be fixed to support IP correctly

We've seen that acl rule installation can take a while as discussed in sonic-net#21988 This check_rule_active/check_rule_counters code was added in sonic-net#21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: sonic-net#22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready

We've seen that acl rule installation can take a while as discussed in sonic-net#21988 This check_rule_active/check_rule_counters code was added in sonic-net#21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: sonic-net#22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready Signed-off-by: Nathan Wolfe <nwolfe@arista.com>

arista-nwolfe · 2026-02-05T19:43:47Z

Just FYI, even with these changes I found tests would sporadically fail with Failed: Acl rule counters are not ready because we weren't waiting long enough so I opened a PR to increase the timeout:
#22260

arista-nwolfe

@mannytaheri are you seeing test_everflow_ipv6.py pass on multi-asic systems with this change?

I'm still seeing all the tests fail with Did not receive expected packet on any of ports and when I look into the state I see all the rules are returning N/A in the counter output:

> aclshow -a
RULE NAME    TABLE NAME      PRIO  PACKETS COUNT    BYTES COUNT
-----------  ------------  ------  ---------------  -------------
RULE_1       EVERFLOWV6      9999  N/A              N/A
RULE_2       EVERFLOWV6      9998  N/A              N/A
RULE_3       EVERFLOWV6      9997  N/A              N/A
RULE_4       EVERFLOWV6      9996  N/A              N/A
RULE_5       EVERFLOWV6      9995  N/A              N/A
RULE_6       EVERFLOWV6      9994  N/A              N/A
RULE_7       EVERFLOWV6      9993  N/A              N/A
RULE_8       EVERFLOWV6      9992  N/A              N/A
RULE_9       EVERFLOWV6      9991  N/A              N/A
RULE_10      EVERFLOWV6      9990  N/A              N/A
RULE_11      EVERFLOWV6      9989  N/A              N/A
RULE_12      EVERFLOWV6      9988  N/A              N/A
RULE_13      EVERFLOWV6      9987  N/A              N/A
RULE_14      EVERFLOWV6      9986  N/A              N/A
RULE_15      EVERFLOWV6      9985  N/A              N/A
RULE_16      EVERFLOWV6      9984  N/A              N/A
RULE_17      EVERFLOWV6      9983  N/A              N/A
RULE_18      EVERFLOWV6      9982  N/A              N/A
RULE_20      EVERFLOWV6      9980  N/A              N/A
RULE_21      EVERFLOWV6      9979  N/A              N/A
RULE_22      EVERFLOWV6      9978  N/A              N/A
RULE_23      EVERFLOWV6      9977  N/A              N/A
RULE_24      EVERFLOWV6      9976  N/A              N/A
RULE_25      EVERFLOWV6      9975  N/A              N/A
RULE_26      EVERFLOWV6      9974  N/A              N/A
rule_999     EVERFLOWV6       999  N/A              N/A
rule_998     EVERFLOWV6       998  N/A              N/A
rule_997     EVERFLOWV6       997  N/A              N/A

Despite all rules showing as active in show acl rule

mannytaheri · 2026-02-09T14:57:31Z

@arista-nwolfe yes, they all pass for me on multi-asic systems with the change.

…for IP_TYPE: ANY Signed-off-by: mtaheri <manouchehr.taheri@nokia.com>

mssonicbld · 2026-02-10T16:30:00Z

/azp run

azure-pipelines · 2026-02-10T16:30:25Z

Azure Pipelines successfully started running 1 pipeline(s).

* Increase timeout for check_rule_active from 60s to 120s We've seen that acl rule installation can take a while as discussed in #21988 This check_rule_active/check_rule_counters code was added in #21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: #22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready Signed-off-by: Nathan Wolfe <nwolfe@arista.com> * Bump up from 120s to 150s as timeouts are still seen Signed-off-by: Nathan Wolfe <nwolfe@arista.com> --------- Signed-off-by: Nathan Wolfe <nwolfe@arista.com>

This is a fix for sonic-mgmt issue sonic-net#21702 (All everflow/test_everflow_ipv6.py are erroring out) This was caused by PR 21384 (Add Support for Egress V6 Testing) Signed-off-by: mtaheri <manouchehr.taheri@nokia.com> Signed-off-by: Zhuohui Tan <zhuohui.tan@amd.com>

…22260) * Increase timeout for check_rule_active from 60s to 120s We've seen that acl rule installation can take a while as discussed in sonic-net#21988 This check_rule_active/check_rule_counters code was added in sonic-net#21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: sonic-net#22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready Signed-off-by: Nathan Wolfe <nwolfe@arista.com> * Bump up from 120s to 150s as timeouts are still seen Signed-off-by: Nathan Wolfe <nwolfe@arista.com> --------- Signed-off-by: Nathan Wolfe <nwolfe@arista.com> Signed-off-by: Zhuohui Tan <zhuohui.tan@amd.com>

This is a fix for sonic-mgmt issue sonic-net#21702 (All everflow/test_everflow_ipv6.py are erroring out) This was caused by PR 21384 (Add Support for Egress V6 Testing) Signed-off-by: mtaheri <manouchehr.taheri@nokia.com> Signed-off-by: Mihut Aronovici <aronovic@cisco.com>

…22260) * Increase timeout for check_rule_active from 60s to 120s We've seen that acl rule installation can take a while as discussed in sonic-net#21988 This check_rule_active/check_rule_counters code was added in sonic-net#21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: sonic-net#22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready Signed-off-by: Nathan Wolfe <nwolfe@arista.com> * Bump up from 120s to 150s as timeouts are still seen Signed-off-by: Nathan Wolfe <nwolfe@arista.com> --------- Signed-off-by: Nathan Wolfe <nwolfe@arista.com> Signed-off-by: Mihut Aronovici <aronovic@cisco.com>

Double commit: sonic-net#22237 During ZTP, the process creates temporary CONFIG_DB settings and if yang validation is enabled, these settings can cause failures. Lets add a YANG model and tests so these won't fail even though these are temporary settings. Failure might look like the below... [342113.096731] sonic-ztp[789615]: main() [342113.096785] sonic-ztp[789615]: File "/usr/local/bin/config_validator.py", line 40, in main 2025 Mar 11 15:43:01.942688 ztp-dut2 INFO sonic-ztp[789615]: main() [342113.096855] sonic-ztp[789615]: raise Exception("Tables without yang models: " + str(yang_parser.tablesWithOutYang)) 2025 Mar 11 15:43:01.942742 ztp-dut2 INFO sonic-ztp[789615]: File "/usr/local/bin/config_validator.py", line 40, in main [342113.096926] sonic-ztp[789615]: Exception: Tables without yang models: {'ZTP': {'mode': {'inband': 'true', 'ipv4': 'true', 'ipv6': 'true', 'out-of-band': 'true', 'product-name': '9716-32D-O-AC-AB-F-US', 'profile': 'active', 'serial-no': '971632D2451040'}}} 2025 Mar 11 15:43:01.942811 ztp-dut2 INFO sonic-ztp[789615]: raise Exception("Tables without yang models: " + str(yang_parser.tablesWithOutYang)) 2025 Mar 11 15:43:01.942882 ztp-dut2 INFO sonic-ztp[789615]: Exception: Tables without yang models: {'ZTP': {'mode': {'inband': 'true', 'ipv4': 'true', 'ipv6': 'true', 'out-of-band': 'true', 'product-name': '9716-32D-O-AC-AB-F-US', 'profile': 'active', 'serial-no': '971632D2451040'}}} 2025 Mar 11 15:43:01.967529 ztp-dut2 INFO sonic-ztp[788029]: sonic_yang(6):Note: Below table(s) have no YANG models: ZTP Why I did it Work item tracking Signed-off-by: rishi rewadkar <rrewadka@cisco.com>

This is a fix for sonic-mgmt issue sonic-net#21702 (All everflow/test_everflow_ipv6.py are erroring out) This was caused by PR 21384 (Add Support for Egress V6 Testing) Signed-off-by: mtaheri <manouchehr.taheri@nokia.com> Signed-off-by: Ravali Yeluri (WIPRO LIMITED) <v-ryeluri@microsoft.com>

…22260) * Increase timeout for check_rule_active from 60s to 120s We've seen that acl rule installation can take a while as discussed in sonic-net#21988 This check_rule_active/check_rule_counters code was added in sonic-net#21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: sonic-net#22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready Signed-off-by: Nathan Wolfe <nwolfe@arista.com> * Bump up from 120s to 150s as timeouts are still seen Signed-off-by: Nathan Wolfe <nwolfe@arista.com> --------- Signed-off-by: Nathan Wolfe <nwolfe@arista.com> Signed-off-by: Ravali Yeluri (WIPRO LIMITED) <v-ryeluri@microsoft.com>

This is a fix for sonic-mgmt issue sonic-net#21702 (All everflow/test_everflow_ipv6.py are erroring out) This was caused by PR 21384 (Add Support for Egress V6 Testing) Signed-off-by: mtaheri <manouchehr.taheri@nokia.com> Signed-off-by: Abhishek <abhishek@nexthop.ai>

…22260) * Increase timeout for check_rule_active from 60s to 120s We've seen that acl rule installation can take a while as discussed in sonic-net#21988 This check_rule_active/check_rule_counters code was added in sonic-net#21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: sonic-net#22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready Signed-off-by: Nathan Wolfe <nwolfe@arista.com> * Bump up from 120s to 150s as timeouts are still seen Signed-off-by: Nathan Wolfe <nwolfe@arista.com> --------- Signed-off-by: Nathan Wolfe <nwolfe@arista.com> Signed-off-by: Abhishek <abhishek@nexthop.ai>

This is a fix for sonic-mgmt issue sonic-net#21702 (All everflow/test_everflow_ipv6.py are erroring out) This was caused by PR 21384 (Add Support for Egress V6 Testing) Signed-off-by: mtaheri <manouchehr.taheri@nokia.com> Signed-off-by: Venkata Gouri Rajesh Etla <vrajeshe@cisco.com>

…22260) * Increase timeout for check_rule_active from 60s to 120s We've seen that acl rule installation can take a while as discussed in sonic-net#21988 This check_rule_active/check_rule_counters code was added in sonic-net#21384 but it never worked on DNX due to unsupported SAI calls: sonic-net/sonic-buildimage#24665 Now that we're working around this issue with: sonic-net#22237 We're starting to run into timeouts waiting for ACL rules: Failed: Acl rule counters are not ready Signed-off-by: Nathan Wolfe <nwolfe@arista.com> * Bump up from 120s to 150s as timeouts are still seen Signed-off-by: Nathan Wolfe <nwolfe@arista.com> --------- Signed-off-by: Nathan Wolfe <nwolfe@arista.com> Signed-off-by: Venkata Gouri Rajesh Etla <vrajeshe@cisco.com>

github-actions bot requested review from StormLiangMS, abdosi and xwjiang-ms February 3, 2026 20:32

[Chassis] Added multi Asics support to everflow

050caff

(cherry picked from commit 654f93c2e6adb01c785f4a18a296a67c590ea157) Signed-off-by: mtaheri <manouchehr.taheri@nokia.com>

mannytaheri force-pushed the everflow-ipv6 branch from 78cc9ac to 050caff Compare February 4, 2026 15:02

mlok-nokia added this to SONiC Chassis Feb 4, 2026

Fixed white space

abe2497

Signed-off-by: mtaheri <manouchehr.taheri@nokia.com>

rlhui added the P0 label Feb 4, 2026

arista-nwolfe reviewed Feb 4, 2026

View reviewed changes

arista-nwolfe mentioned this pull request Feb 5, 2026

Increase timeout for check_rule_active from 60s to 120s #22260

Merged

12 tasks

arista-nwolfe reviewed Feb 6, 2026

View reviewed changes

Modified IP_TYPE: IPV6ANY to IP_TYPE: ANY since it brm added support …

c9b4aff

…for IP_TYPE: ANY Signed-off-by: mtaheri <manouchehr.taheri@nokia.com>

judyjoseph requested a review from arlakshm February 11, 2026 18:20

abdosi approved these changes Feb 18, 2026

View reviewed changes

abdosi merged commit 0dd38f6 into sonic-net:master Feb 18, 2026
18 checks passed

github-project-automation bot moved this to Done in SONiC Chassis Feb 18, 2026

rlhui mentioned this pull request Feb 25, 2026

Bug: [Chassis] All everflow/test_everflow_ipv6.py are erroring ou #21702

Closed

bpar9 mentioned this pull request Mar 4, 2026

xfail test_everflow_ipv6 for known issue for now #22188

Closed

12 tasks

Conversation

mannytaheri commented Feb 3, 2026

Description of PR

Type of change

Back port request

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

Uh oh!

mssonicbld commented Feb 3, 2026

Uh oh!

azure-pipelines bot commented Feb 3, 2026

Uh oh!

mssonicbld commented Feb 4, 2026

Uh oh!

azure-pipelines bot commented Feb 4, 2026

Uh oh!

mssonicbld commented Feb 4, 2026

Uh oh!

azure-pipelines bot commented Feb 4, 2026

Uh oh!

arista-nwolfe Feb 4, 2026

Choose a reason for hiding this comment

Uh oh!

mannytaheri Feb 4, 2026

Choose a reason for hiding this comment

Uh oh!

arista-nwolfe Feb 4, 2026

Choose a reason for hiding this comment

Uh oh!

mannytaheri Feb 4, 2026

Choose a reason for hiding this comment

Uh oh!

eswaran-nexthop Feb 4, 2026

Choose a reason for hiding this comment

Uh oh!

eswaran-nexthop Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

eswaran-nexthop Feb 5, 2026

Choose a reason for hiding this comment

Uh oh!

arista-nwolfe commented Feb 5, 2026

Uh oh!

arista-nwolfe left a comment

Choose a reason for hiding this comment

Uh oh!

mannytaheri commented Feb 9, 2026

Uh oh!

mssonicbld commented Feb 10, 2026

Uh oh!

azure-pipelines bot commented Feb 10, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants