[featured] fix non existing feature start

[stepanblyschak]

Fix an issue that when starting a non existing feature and then installing it with sonic-package-manager it won't start.

E.g:

admin@sonic:~$ redis-cli -n 4 hmset 'FEATURE|non-existing' state enabled OK
admin@sonic:~$ sudo zless /var/log/syslog | grep featured 2025 Mar 26 15:19:33.237090 sonic INFO featured: Feature non-existing is enabled and started

With this patch an error is recordded and feature state is not cached:

2025 Mar 26 15:35:32.794640 sonic INFO featured: Running cmd: '['sudo', 'systemctl', 'unmask', 'non-existing.service']'
2025 Mar 26 15:35:33.233685 sonic INFO featured: Output:  , Stderr: Unit non-existing.service does not exist, proceeding anyway.
2025 Mar 26 15:35:33.233743 sonic INFO featured: Running cmd: '['sudo', 'systemctl', 'enable', 'non-existing.service']'
2025 Mar 26 15:35:33.247238 sonic ERR featured: ['sudo', 'systemctl', 'enable', 'non-existing.service'] - failed: return code - 1, output:
2025 Mar 26 15:35:33.247300 sonic ERR featured: Feature 'non-existing.service' failed to be enabled and started
2025 Mar 26 15:35:39.331762 sonic INFO featured: Running cmd: '['sudo', 'systemctl', 'unmask', 'non-existing.service']'
2025 Mar 26 15:35:39.780474 sonic INFO featured: Output:  , Stderr: Unit non-existing.service does not exist, proceeding anyway.
2025 Mar 26 15:35:39.780513 sonic INFO featured: Running cmd: '['sudo', 'systemctl', 'enable', 'non-existing.service']'
2025 Mar 26 15:35:39.795113 sonic ERR featured: ['sudo', 'systemctl', 'enable', 'non-existing.service'] - failed: return code - 1, output:
2025 Mar 26 15:35:39.795153 sonic ERR featured: Feature 'non-existing.service' failed to be enabled and started

Feature will start next time when setting state field in FEATURE table

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[yejianquan]

Hi @stepanblyschak , this change is blocking submodule advance in master and 202505 branch, could you check:
master:
sonic-net/sonic-buildimage#22160
https://elastictest.org/scheduler/testplan/68451b5304595a8ba1c1f6e3?testcase=pc%2Ftest_lag_member_forwarding.py%7C%7C%7C2&type=console

202505:
sonic-net/sonic-buildimage#22660
https://elastictest.org/scheduler/testplan/6843da51d2f1accb40ff9130?testcase=pc%2Ftest_retry_count.py%7C%7C%7C2&type=console

@qiluo-msft @zbud-msft for viz, since in 202505, telemetry is the only one reporting issue(in master, both frr_bmp and telemetry report issues), not sure whether it's the hidden issue is exposed.

[zbud-msft]

Hi @yejianquan, telemetry is not a built container in public image. Thats why test fails when running sudo systemctl stop telemetry.service.

[yejianquan]

Hi @yejianquan, telemetry is not a built container in public image. Thats why test fails when running sudo systemctl stop telemetry.service.

Hi @zbud-msft so we should add the error msg into common ignore list of log analyzer?

[zbud-msft]

Hi @stepanblyschak seems like featured is trying to stop telemetry.service that is already disabled in public image. Seems like this PR is responsible for that change and needs to be fixed.

[stepanblyschak]

@zbud-msft Featured does not do anything special with telemetry.

Errors:

E               Match Messages:
E               2025 Jun  8 06:54:56.883029 vlab-03 ERR featured: ['sudo', 'systemctl', 'enable', 'frr_bmp.service'] - failed: return code - 1, output:
E               
E               2025 Jun  8 06:54:56.883076 vlab-03 ERR featured: Feature 'frr_bmp.service' failed to be enabled and started
E               
E               2025 Jun  8 06:55:11.159750 vlab-03 ERR featured: ['sudo', 'systemctl', 'stop', 'telemetry.service'] - failed: return code - 5, output:
E               
E               2025 Jun  8 06:55:11.159991 vlab-03 ERR featured: Feature 'telemetry.service' failed to be stopped and disabled

could happen only if user provided telemetry in CONFIG_DB when telemetry is not built in the image and correctly reports the issue.