Bug: [NTP] IPv6 synchronization tests fail with “No suitable source for synchronisation”

[Janetxxx]

Is it platform specific

Mellanox

Importance or Severity

Critical

Description of the bug

ntp/test_ntp.py fails in the IPv6 variants on Mellanox-SN2700-A1 / T0.
The logs show that the PTF uses ntpsec with interface listen mgmt, without an explicit IPv6 listen directive.

Steps to Reproduce

Run ntp/test_ntp.py, including the ipv6_only variants. Observe synchronization failures when the DUT uses the PTF’s management IPv6 address as its NTP server.

Actual Behavior and Expected Behavior

test_ntp_long_jump_enabled[ipv6_only] fails after waiting 720 seconds for synchronization.
test_ntp[ipv6_only] fails because chronyd -q exits with code 1.

Relevant log output

if (hostname_res.is_failed or 'exception' in hostname_res) and not module_ignore_errors:
>           raise RunAnsibleModuleFail("run module {} failed".format(module_name), hostname_res)
E           tests.common.errors.RunAnsibleModuleFail: run module command failed, Ansible Results =>
E           failed = True
E           changed = True
E           rc = 1
E           cmd = ['timeout', '20', 'chronyd', '-q', '-F', '1']
E           start = 2026-09-19 03:53:50.834822
E           end = 2026-09-19 03:54:01.204556
E           delta = 0:00:10.369734
E           msg = non-zero return code
E           invocation = {'module_args': {'_raw_params': 'timeout 20 chronyd -q -F 1', '_uses_shell': False, 'expand_argument_vars': True, 'stdin_add_newline': True, 'strip_empty_ends': True, 'cmd': None, 'argv': None, 'chdir': None, 'executable': None, 'creates': None, 'removes': None, 'stdin': None}}
E           exception = (traceback unavailable)
E           _ansible_no_log = False
E           stdout =
E           stderr =
E           2026-09-19T03:53:50Z chronyd version 4.6.1 starting (+CMDMON +NTP +REFCLOCK +RTC +PRIVDROP +SCFILTER +SIGND +ASYNCDNS +NTS +SECHASH +IPV6 -DEBUG)
E           2026-09-19T03:53:50Z Using right/UTC timezone to obtain leap second data
E           2026-09-19T03:53:50Z Frequency -12.464 +/- 0.061 ppm read from /var/lib/chrony/chrony.drift
E           2026-09-19T03:53:50Z Loaded seccomp filter (level 1)
E           2026-09-19T03:54:01Z No suitable source for synchronisation
E           2026-09-19T03:54:01Z chronyd exiting

=========================== short test summary info ============================
FAILED ntp/test_ntp.py::test_ntp_long_jump_enabled[ipv6_only] - Failed: NTP l...
FAILED ntp/test_ntp.py::test_ntp[ipv6_only] - tests.common.errors.RunAnsibleM...

Output of show version

SONiC 20260510.16

Attach files (if any)

No response

[Janetxxx]

Root cause identified: the PTF restart flow renumber_topo.yml was missing default.accept_ra=0, and neither the add-topo nor restart flow explicitly disabled RA on the mgmt interface.
After PTF recreation, mgmt.accept_ra remained 1. The PTF learned an unwanted IPv6 /64 on-link route, causing it to attempt direct neighbor discovery for the DUT instead of forwarding through the configured gateway. This prevented IPv6 NTP synchronization.

[Janetxxx]

Cherry-picked #25065, which adds the missing default RA setting to the restart flow and explicitly sets mgmt.accept_ra=0 in both flows. Verified it in 202605 nightly run.