Is it platform specific
generic
Importance or Severity
Critical
Description of the bug
Summary
test_snmp_interfaces intermittently fails with a missing 'name' field for the VLAN interface returned by SNMP. This appears to be a race condition where SNMP responds before all MIB data is fully populated.
Issue
The current wait=True logic in get_snmp_facts() only waits for the subagent to start but does not guarantee that all MIB fields are ready. Even though the subagent is running, the VLAN MIB fields may need slightly more time before querying.
sonic-mgmt/tests/common/helpers/snmp_helpers.py
def get_snmp_facts(duthost, localhost, host, version, community, is_dell=False, module_ignore_errors=False,
wait=False, include_swap=False, timeout=DEF_WAIT_TIMEOUT, interval=DEF_CHECK_INTERVAL,
snmp_timeout=SNMP_DEFAULT_TIMEOUT):
if not wait:
return _get_snmp_facts(localhost, host, version, community, is_dell, include_swap, module_ignore_errors,
timeout=snmp_timeout)
global global_snmp_facts
pytest_assert(wait_until(timeout, interval, 0, _update_snmp_facts, localhost, host, version,
community, is_dell, include_swap, duthost, snmp_timeout), "Timeout waiting for SNMP facts")
return global_snmp_facts
A similar issue was reported for test_snmp_fdb.py and was fixed by introducing a longer wait_until.
Log snippet showing missing 3000 'name'
23/12/2025 13:19:08 base._run L0073 DEBUG | /var/AzDevOps/tests/common/helpers/snmp_helpers.py::_get_snmp_facts#30: [localhost] AnsibleModule::snmp_facts, args=[], kwargs={"host": "192.168.121.10", "version": "v2c", "community": "public", "is_dell": false, "module_ignore_errors": false, "include_swap": false, "timeout": 20}
23/12/2025 13:19:18 base._run L0111 DEBUG | /var/AzDevOps/tests/common/helpers/snmp_helpers.py::_get_snmp_facts#30: [localhost] AnsibleModule::snmp_facts Result => ...
"snmp_interfaces": {
"1": {"ifindex": "0", "name": "etp1a", ...},
...
"10000": {"ifindex": "9999", "name": "eth0", ...},
"3000": {"ifHighSpeed": "40000", "description": ""}
}
Example CLI output for snmp get
Healthy
~% snmpget -v2c -c public 192.168.121.10 1.3.6.1.2.1.2.2.1.2.3000
IF-MIB::ifDescr.3000 = STRING: Vlan1000
During config reload
~% snmpget -v2c -c public 192.168.121.10 1.3.6.1.2.1.2.2.1.2.3000
IF-MIB::ifDescr.3000 = No Such Object available on this agent at this OID
Steps to Reproduce
test_snmp_interfaces failed with the KeyError: 'name' issue 2/13 time when running tests in this order.
snmp/test_snmp_cpu.py::test_snmp_cpu
snmp/test_snmp_default_route.py::test_snmp_default_route
snmp/test_snmp_fdb.py::test_snmp_fdb_send_tagged
snmp/test_snmp_interfaces.py::test_snmp_interfaces
Actual Behavior and Expected Behavior
Before tests run: full data returned
During/just after config reload: missing key 'name' for specific interfaces (e.g., ifIndex 3000)
- Expected is all MIB data is returned (including 'name')
Relevant log output
localhost = <tests.common.devices.local.Localhost object at 0x7f765d034760>
creds_all_duts = {'upscaleai-t0': {'7nodes_cisco_P1': 'vars/configdb_jsons/7nodes_cisco/P1.json', '7nodes_cisco_P2': 'vars/config...o_P3': 'vars/configdb_jsons/7nodes_cisco/P3.json', '7nodes_cisco_P4': 'vars/configdb_jsons/7nodes_cisco/P4.json', ...}}
duthosts = [<MultiAsicSonicHost upscaleai-t0>]
enum_rand_one_per_hwsku_hostname = 'upscaleai-t0'
@pytest.mark.bsl
def test_snmp_interfaces(localhost, creds_all_duts, duthosts, enum_rand_one_per_hwsku_hostname):
"""compare the snmp facts between observed states and target state"""
duthost = duthosts[enum_rand_one_per_hwsku_hostname]
hostip = duthost.host.options['inventory_manager'].get_host(
duthost.hostname).vars['ansible_host']
snmp_facts = get_snmp_facts(
duthost, localhost, host=hostip, version="v2c",
community=creds_all_duts[duthost.hostname]["snmp_rocommunity"], wait=True)['ansible_facts']
> snmp_ifnames = [v['name']
for k, v in list(snmp_facts['snmp_interfaces'].items())]
creds_all_duts = {'t0': {'7nodes_cisco_P1': 'vars/configdb_jsons/7nodes_cisco/P1.json', '7nodes_cisco_P2': 'vars/config...o_P3': 'vars/configdb_jsons/7nodes_cisco/P3.json', '7nodes_cisco_P4': 'vars/configdb_jsons/7nodes_cisco/P4.json', ...}}
duthost = <MultiAsicSonicHost upscaleai-t0>
duthosts = [<MultiAsicSonicHost upscaleai-t0>]
enum_rand_one_per_hwsku_hostname = 'upscaleai-t0'
hostip = '192.168.121.10'
localhost = <tests.common.devices.local.Localhost object at 0x7f765d034760>
snmp_facts = {'ansible_all_ipv4_addresses': [], 'ansible_sysCachedMemory': 12235164, 'ansible_sysTotalBuffMemory': 99132, 'ansible_sysTotalFreeMemory': 14617100, ...}
snmp/test_snmp_interfaces.py:171:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.0 = <list_iterator object at 0x7f764fb56d60>
> snmp_ifnames = [v['name']
for k, v in list(snmp_facts['snmp_interfaces'].items())]
E KeyError: 'name'
.0 = <list_iterator object at 0x7f764fb56d60>
k = '3000'
v = {'description': '', 'ifHighSpeed': '40000'}
snmp/test_snmp_interfaces.py:171: KeyErrorOutput of show version
Attach files (if any)
No response
Is it platform specific
generic
Importance or Severity
Critical
Description of the bug
Summary
test_snmp_interfaces intermittently fails with a missing 'name' field for the VLAN interface returned by SNMP. This appears to be a race condition where SNMP responds before all MIB data is fully populated.
Issue
The current wait=True logic in
get_snmp_facts()only waits for the subagent to start but does not guarantee that all MIB fields are ready. Even though the subagent is running, the VLAN MIB fields may need slightly more time before querying.sonic-mgmt/tests/common/helpers/snmp_helpers.py
A similar issue was reported for
test_snmp_fdb.pyand was fixed by introducing a longerwait_until.Log snippet showing missing 3000 'name'
Example CLI output for snmp get
Healthy
During config reload
Steps to Reproduce
test_snmp_interfaces failed with the
KeyError: 'name'issue 2/13 time when running tests in this order.snmp/test_snmp_cpu.py::test_snmp_cpu
snmp/test_snmp_default_route.py::test_snmp_default_route
snmp/test_snmp_fdb.py::test_snmp_fdb_send_tagged
snmp/test_snmp_interfaces.py::test_snmp_interfaces
Actual Behavior and Expected Behavior
Before tests run: full data returned
During/just after config reload: missing key 'name' for specific interfaces (e.g., ifIndex 3000)
Relevant log output
Output of
show versionAttach files (if any)
No response