Link event damping feature by sivat6 · Pull Request #1906 · sonic-net/sonic-sairedis

sivat6 · 2026-05-22T11:30:18Z

PR#1331 is not yet merged. Added the changes of Add libsai RedisInterface for link event damping. #1331.

Other changes:

Feature Functionality implementation was missing:
The existing implementation only defines structures and enums for configuration provisioning. The damping functional logic was not implemented. Implemented the damping algorithm as mentioned in the HLD and invoked during the link event handling.
Missing Damping Statistics
Damping-related counters mentioned in HLD are not present in STATS_DB. Added the damping stats (pre & post damping counters) to STATS_DB
Physical vs Operational State Mismatch
The damping algorithm is triggered during link up/down events, with timer and threshold checks handled there.
Once the maximum suppress time is reached, the actual status update occurs only during the next link event, causing a mismatch. Added the timer thread which periodically checks the damping timer and if the damping ends, it updates the physical link status.

Test Plan: Test plan is already in place (https://github.com/sivat6/sonic-mgmt/blob/master/docs/testplan/Link_event_damping.md)

Test Results: This test plan has 9 major test cases and there are about 54 total test cases (Including sub test cases). All these 54 test cases are executed.

mssonicbld · 2026-05-22T11:30:27Z

/azp run

azure-pipelines · 2026-05-22T11:30:40Z

Azure Pipelines successfully started running 1 pipeline(s).

prsunny · 2026-05-22T19:01:33Z

@sivat6 , can you check the build failure?

prsunny · 2026-05-22T19:06:04Z

HLD PR - sonic-net/SONiC#1071

Copilot

Pull request overview

This PR implements the link event damping feature end-to-end in sairedis/syncd: it introduces Redis-extension PORT attributes to configure damping, applies the damping algorithm to port state change notifications in syncd, and adds a periodic timer to end suppression even without new link events.

Changes:

Add Redis-extension PORT attributes and plumbing (libsai ⇄ syncd) for link-event damping algorithm/config.
Apply damping during port state change notification processing, plus a timer thread to enforce max-suppress-time expiry.
Emit per-port damping counters/state into a dedicated DB table for observability.

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 10 comments.

Show a summary per file

File	Description
`unittest/lib/TestClientServerSai.cpp`	Adds unit tests for setting Redis PORT damping attributes and client-mode rejection.
`syncd/tests/TestSyncdLinkEventDamping.cpp`	Adds a syncd integration-style test for damping config set behavior.
`syncd/tests/Makefile.am`	Registers the new syncd damping test source file.
`syncd/Syncd.h`	Introduces per-port damping state structure, helper APIs, and thread/db members.
`syncd/Syncd.cpp`	Implements config handling, damping algorithm, timer thread, and stats publishing.
`syncd/NotificationProcessor.h`	Extends constructor to accept an optional “apply damping” callback.
`syncd/NotificationProcessor.cpp`	Filters port state change notifications by applying damping before sending.
`lib/sairediscommon.h`	Adds the Redis op name for damping config set.
`lib/Sai.cpp`	Bypasses metadata for Redis PORT extension attributes and routes them to RedisSai.
`lib/RedisRemoteSaiInterface.h`	Declares Redis PORT attribute detection and damping config setter helpers.
`lib/RedisRemoteSaiInterface.cpp`	Implements Redis PORT attribute set handling and op dispatch to syncd.
`lib/ClientSai.cpp`	Rejects Redis extension attrs (including Redis PORT attrs) in CLIENT mode.

prsunny · 2026-05-22T19:10:57Z

Please address copilot comments

mssonicbld · 2026-05-27T04:58:58Z

/azp run

azure-pipelines · 2026-05-27T04:59:09Z

Azure Pipelines successfully started running 1 pipeline(s).

sivat6 · 2026-05-27T05:05:43Z

Please address copilot comments

Addressed all copilot comments.

mssonicbld · 2026-05-27T05:06:12Z

/azp run

azure-pipelines · 2026-05-27T05:06:23Z

Azure Pipelines successfully started running 1 pipeline(s).

mssonicbld · 2026-05-27T09:54:59Z

/azp run

azure-pipelines · 2026-05-27T09:55:12Z

Azure Pipelines successfully started running 1 pipeline(s).

mssonicbld · 2026-05-27T10:33:58Z

/azp run

azure-pipelines · 2026-05-27T10:34:11Z

Azure Pipelines successfully started running 1 pipeline(s).

prsunny · 2026-05-28T00:09:09Z

Can you check the build failure?

mssonicbld · 2026-05-28T04:02:58Z

/azp run

azure-pipelines · 2026-05-28T04:03:10Z

Azure Pipelines successfully started running 1 pipeline(s).

mssonicbld · 2026-05-28T05:30:12Z

/azp run

azure-pipelines · 2026-05-28T05:30:23Z

Azure Pipelines successfully started running 1 pipeline(s).

mssonicbld · 2026-05-28T06:53:42Z

/azp run

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

…yncd/tests/TestSyncdLinkEventDamping.cpp Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

mssonicbld · 2026-06-12T04:49:31Z

/azp run

azure-pipelines · 2026-06-12T04:49:43Z

Azure Pipelines successfully started running 1 pipeline(s).

mssonicbld · 2026-06-12T08:22:36Z

Hi, there are workflow run(s) waiting for approval, you may be first-time contributor. I will notify maintainers to help approve once PR is approved. Thanks!

_{---Powered by SONiC BuildBot}

sivat6 · 2026-06-12T09:09:40Z

/azpw retry

mssonicbld · 2026-06-12T09:09:43Z

Retrying failed(or canceled) jobs...

mssonicbld · 2026-06-12T09:09:46Z

Retrying failed(or canceled) stages in build 1136667:

✅Stage Test:

Job vstest: retried.

kcudnik · 2026-06-20T07:00:15Z

+    /**
+     * @brief Link event damping configuration and state per port
+     */
+    struct LinkEventDampingPortState


this should go to separate header

kcudnik · 2026-06-20T07:00:27Z

+             * @return true if should suppress, false if should propagate
+             */
+            bool applyAiedAlgorithm(
+		    _In_ sai_object_id_t portVid,


wrong alignment

kcudnik · 2026-06-20T07:01:58Z

+    // Damping exits when EITHER:
+    // 1. Time-based: damping_duration_ms >= max_suppress_time
+    // 2. Penalty-based: current_penalty < reuse_threshold (decay-based recovery)
+    if (state.is_damping_active)
+    {
+        // Check timeout - never suppress longer than max_suppress_time
+        // This is a hard timestamp-based limit to prevent infinite suppression
+        uint64_t damping_duration_ms = currentTimeMs - state.damping_start_time_ms;
+
+        if (damping_duration_ms >= state.aied_config.max_suppress_time)
+        {
+            // Store temporary strings to avoid dangling pointers
+            std::string physicalStatusStr = sai_serialize_port_oper_status(state.physical_status);
+            std::string advertisedStatusStr = sai_serialize_port_oper_status(state.advertised_status);
+            SWSS_LOG_NOTICE("Port VID %s exiting damped state: max suppress time (%u ms) "
+                    "exceeded. Duration: %lu ms. Physical state: %s, Advertised state: %s",
+                    portVidStr.c_str(), state.aied_config.max_suppress_time,
+                    damping_duration_ms, physicalStatusStr.c_str(),
+                    advertisedStatusStr.c_str());
+            state.is_damping_active = false;
+            state.damping_start_time_ms = 0;  // Reset timer when exiting damping
+
+            // Propagate last link event when penalty decays below reuse threshold
+            if (state.advertised_status != state.physical_status)
+            {
+                state.pending_state_sync = true;
+                SWSS_LOG_NOTICE("Port VID %s state mismatch detected on damping "
+                        "exit (timeout): physical=%s, advertised=%s. "
+                        "Marking for state sync on next notification.",
+                        portVidStr.c_str(), physicalStatusStr.c_str(),
+                        advertisedStatusStr.c_str());
+            }
+            state.advertised_status = state.physical_status;
+
+            // Write updated state to STATE_DB after exiting damping
+            writeDampingCountersToStateDb(portVid, state);
+        }
+        // Check reuse threshold - exit if penalty decays below threshold
+        // Penalty decays based on last_decay_time tracking
+        else if (state.current_penalty < state.aied_config.reuse_threshold)
+        {
+            // Store temporary strings to avoid dangling pointers
+            std::string physicalStatusStr = sai_serialize_port_oper_status(state.physical_status);
+            std::string advertisedStatusStr = sai_serialize_port_oper_status(state.advertised_status);
+            SWSS_LOG_NOTICE("Port VID %s exiting damped state: penalty (%u) < "
+                    "reuse_threshold (%u). Penalty decayed due to exponential decay "
+                    "formula. Physical state: %s, Advertised state: %s",
+                    portVidStr.c_str(), state.current_penalty,
+                    state.aied_config.reuse_threshold, physicalStatusStr.c_str(),
+                    advertisedStatusStr.c_str());
+            state.is_damping_active = false;


code quality in this function and ohters really suffers, and its not keeping same standard as entire project missing spaces empty lines, just many blobs of code not readable, commnets before else if ..

prsunny reviewed May 22, 2026

View reviewed changes

Comment thread lib/ClientSai.cpp

prsunny requested a review from Copilot May 22, 2026 19:04

Copilot started reviewing on behalf of prsunny May 22, 2026 19:04 View session

prsunny requested review from saiarcot895 and stephenxs May 22, 2026 19:04

Copilot AI reviewed May 22, 2026

View reviewed changes

github-advanced-security AI found potential problems May 22, 2026

View reviewed changes

Comment thread syncd/Syncd.cpp Fixed

Comment thread syncd/Syncd.cpp Fixed

sivat6 closed this May 27, 2026

sivat6 reopened this May 27, 2026

sivat6 force-pushed the link_event_damping_feature branch from 4bb4573 to 8f4866a Compare May 28, 2026 04:02

sivat6 added 15 commits June 12, 2026 10:18

Test purpose to see PR# 1925 fixes current build issue

479f546

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Remove the code of PR# 1925

a17d402

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Fixing build issue

e162b57

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

SWSS_LOG_ENTER missed in new API

1b2802c

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Adding suppress and reuse threshold invalid config check

9a037f5

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Adding tests to increase coverage

564e2b1

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Added SWSS_LOG_ENTER() in the APIs in test file

a864f51

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Addressed review comments

0e31ec1

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Addressed code review comments

7401665

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Adding tests under unittest and removing the unnecessary tests from s…

030fa7d

…yncd/tests/TestSyncdLinkEventDamping.cpp Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Trigger pipeline

0cce42b

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Adding SWSS_LOG_ENTER()

8a2b883

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Debugging the test case

c59f614

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Tweaking some timing value in one test

55d92c3

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

Tweaking params

6fc0dd4

Signed-off-by: Sivakumar Thirukkanna Thevar <sthirukkanna@juniper.net>

sivat6 force-pushed the link_event_damping_feature branch from 46ac713 to 6fc0dd4 Compare June 12, 2026 04:49

prsunny approved these changes Jun 12, 2026

View reviewed changes

mramezani95 merged commit a64c0fe into sonic-net:master Jun 12, 2026
19 checks passed

yijingyan2 mentioned this pull request Jun 15, 2026

[submodule][master] Update submodule sonic-sairedis to the latest HEAD automatically sonic-net/sonic-buildimage#27870

Open

DendroLabs mentioned this pull request Jun 20, 2026

[Link Event Damping] Add per-port link event damper with syslog and monitor-only mode #1936

Closed

kcudnik reviewed Jun 20, 2026

View reviewed changes

lolyu mentioned this pull request Jun 21, 2026

Fixing sonic-buildimage failures due to PR#1906 #1942

Open

1 task

Conversation

sivat6 commented May 22, 2026

Uh oh!

mssonicbld commented May 22, 2026

Uh oh!

azure-pipelines Bot commented May 22, 2026

Uh oh!

Uh oh!

prsunny commented May 22, 2026

Uh oh!

prsunny commented May 22, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

prsunny commented May 22, 2026

Uh oh!

Uh oh!

Uh oh!

mssonicbld commented May 27, 2026

Uh oh!

azure-pipelines Bot commented May 27, 2026

Uh oh!

sivat6 commented May 27, 2026

Uh oh!

mssonicbld commented May 27, 2026

Uh oh!

azure-pipelines Bot commented May 27, 2026

Uh oh!

mssonicbld commented May 27, 2026

Uh oh!

azure-pipelines Bot commented May 27, 2026

Uh oh!

mssonicbld commented May 27, 2026

Uh oh!

azure-pipelines Bot commented May 27, 2026

Uh oh!

prsunny commented May 28, 2026

Uh oh!

mssonicbld commented May 28, 2026

Uh oh!

azure-pipelines Bot commented May 28, 2026

Uh oh!

mssonicbld commented May 28, 2026

Uh oh!

azure-pipelines Bot commented May 28, 2026

Uh oh!

mssonicbld commented May 28, 2026

Uh oh!

mssonicbld commented Jun 12, 2026

Uh oh!

azure-pipelines Bot commented Jun 12, 2026

Uh oh!

mssonicbld commented Jun 12, 2026

Uh oh!

sivat6 commented Jun 12, 2026

Uh oh!

mssonicbld commented Jun 12, 2026

Uh oh!

mssonicbld commented Jun 12, 2026

Uh oh!

Uh oh!

kcudnik Jun 20, 2026

Choose a reason for hiding this comment

Uh oh!

kcudnik Jun 20, 2026

Choose a reason for hiding this comment

Uh oh!