[swss] Restart countersyncd after unexpected exit

[Pterosaur]

Why I did it

swss:countersyncd is currently started through dependent-startup, but its supervisor stanza uses autorestart=false.
As a result, when countersyncd exits unexpectedly, supervisor leaves it in EXITED instead of recovering it.
That behavior was reproduced on str4-sn5640-2, where HFT telemetry validation could pass while countersyncd still stayed down afterward.

Work item tracking

• Microsoft ADO (number only): N/A

How I did it

• change countersyncd from autorestart=false to autorestart=unexpected
• add startsecs=10
• add startretries=3

This keeps the existing dependent-startup flow, allows supervisor to recover countersyncd after unexpected exits, and gives startretries a meaningful guardrail for short-lived crash loops before a process is considered stably started.

How to verify it

1. Apply the countersyncd supervisor stanza change on the DUT swss container.
2. Run:

   ./run_tests.sh -u -n vms70-t0-sn5640-2 -i ../ansible/str4,../ansible/veos -l info -m individual -e --skip_sanity -e --disable_loganalyzer -c "high_frequency_telemetry/test_high_frequency_telemetry.py::test_hft_port_counters"

3. Confirm:
   • HFT telemetry validation still passes
   • docker exec swss supervisorctl status countersyncd reports RUNNING after the test

Observed result from live DUT validation:

• DUT: str4-sn5640-2
• Platform: x86_64-nvidia_sn5640-r0 / Mellanox-SN5640-C512S2
• Image: 20241212.54
• Test result: 1 passed

Which release branch to backport (provide reason below if selected)

• 202305
• 202311
• 202405
• 202411
• 202505
• 202511

Backport reason:

• This is a bug fix for service recovery behavior in swss.
• The same PR is also requested for msft-202412 via label because unexpected countersyncd exits leave the supervisor-managed service down until manual intervention.

Tested branch (Please provide the tested image version)

• 20241212.54 (live-patched DUT validation on str4-sn5640-2)

Description for the changelog

Restart countersyncd automatically after unexpected exits in swss.

Link to config_db schema for YANG module changes

N/A

A picture of a cute animal (not mandatory but encouraged)

🦦

[Copilot]

Pull request overview

Adjusts the swss container’s supervisor configuration so countersyncd can recover automatically after unexpected exits, improving telemetry robustness without changing the dependent-startup ordering.

Changes:

• Switch countersyncd from autorestart=false to autorestart=unexpected
• Add startsecs=1 and startretries=3 to constrain restart behavior for immediate start failures

---

You can also share your feedback on Copilot code review. Take the survey.

[banidoru]

Clean, well-scoped fix. The three supervisor settings work correctly together:

• autorestart=unexpected recovers countersyncd after non-zero exits without interfering with intentional stops.
• startsecs=10 ensures short-lived crash loops consume startretries rather than being treated as successful-then-crashed (which would restart indefinitely).
• startretries=3 caps recovery attempts before supervisor gives up, preventing infinite restart loops.

The earlier feedback on startsecs has been addressed. No concerns with this change.

[banidoru]

Clean, minimal fix. The three supervisor knobs (autorestart=unexpected, startsecs=10, startretries=3) work correctly together: crashes within the first 10 s consume a retry (capped at 3), while crashes after 10 s trigger an immediate restart without burning retries. This matches the stated goal of recovering from unexpected exits without infinite crash loops. The prior review feedback on startsecs has been addressed. No concerns.

[banidoru]

Clean, minimal fix. Changing autorestart=false → autorestart=unexpected correctly allows supervisord to recover countersyncd after unexpected exits while still respecting intentional stops. startsecs=10 (updated from the original 1s per review feedback) ensures that short-lived crash loops consume the 3 retries before supervisor gives up, preventing infinite restart storms. No concerns with correctness, security, or design.

[banidoru]

All reviewers approved. LGTM.

[Pterosaur]

/azpw run

[mssonicbld]

/AzurePipelines run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[Pterosaur]

Created 202412 backport PR: Azure/sonic-buildimage-msft#2056

[kperumalbfn]

@Pterosaur what are the reasons for unexpected exit of countersyncd??

[mssonicbld]

Cherry-pick PR to msft-202412:

[Pterosaur]

@Pterosaur what are the reasons for unexpected exit of countersyncd??

Hi @kperumalbfn , if the downstream service, such as otel collector, was not started or exited. We will proactively terminate countersyncd to drain the counters data.