UnavailableDataError for Key "COUNTERS_PORT_NAME_MAP" in COUNTERS_DB about sonic-snmpagent

[sripalnati]

Hi, the above unavailableDataError is always seen in the syslog when swss is restarted. But, when I connect to the redis-cli and I could able to dump the COUNTERS_PORT_NAME_MAP as follows:

127.0.0.1:6379[2]> hgetall COUNTERS_PORT_NAME_MAP

1. "Ethernet0"
2. "oid:0x100000000049a"
3. "Ethernet4"
4. "oid:0x100000000049b"
5. "Ethernet8"
6. "oid:0x100000000049c"
7. "Ethernet12"
8. "oid:0x100000000049d"
9. "Ethernet16"
10. "oid:0x100000000049e"
11. "Ethernet20"
12. "oid:0x100000000049f"
13. "Ethernet24"
14. "oid:0x10000000004a0"
15. "Ethernet28"
16. "oid:0x10000000004a1"
17. "Ethernet32"
18. "oid:0x10000000004a2"
19. "Ethernet36"
20. "oid:0x10000000004a3"
21. "Ethernet40"
22. "oid:0x10000000004a4"
23. "Ethernet44"
24. "oid:0x10000000004a5"
25. "Ethernet48"
26. "oid:0x10000000004a6"
27. "Ethernet52"
28. "oid:0x10000000004a7"
29. "Ethernet56"
30. "oid:0x10000000004a8"
31. "Ethernet60"
32. "oid:0x10000000004a9"
33. "Ethernet64"
34. "oid:0x10000000004aa"
35. "Ethernet68"
36. "oid:0x10000000004ab"
37. "Ethernet72"
38. "oid:0x1000000000488"
39. "Ethernet76"
40. "oid:0x1000000000489"
41. "Ethernet80"
42. "oid:0x100000000048a"
43. "Ethernet84"
44. "oid:0x100000000048b"
45. "Ethernet88"
46. "oid:0x100000000048c"
47. "Ethernet92"
48. "oid:0x100000000048d"
49. "Ethernet96"
50. "oid:0x100000000048e"
51. "Ethernet100"
52. "oid:0x100000000048f"
53. "Ethernet104"
54. "oid:0x1000000000490"
55. "Ethernet108"
56. "oid:0x1000000000491"
57. "Ethernet112"
58. "oid:0x1000000000492"
59. "Ethernet116"
60. "oid:0x1000000000493"
61. "Ethernet120"
62. "oid:0x1000000000494"
63. "Ethernet124"
64. "oid:0x1000000000495"
65. "Ethernet128"
66. "oid:0x1000000000496"
67. "Ethernet132"
68. "oid:0x1000000000497"
69. "Ethernet136"
70. "oid:0x1000000000498"
71. "Ethernet140"
72. "oid:0x1000000000499"
    127.0.0.1:6379[2]>

[jerry-chang3300]

We have similar error log as previous, it seems that SNMP tries to get queue_stat_map before it's created. Since SNMP would retry to get it again, it should not be a problem, so we low down the log level from error to warning.

"Oct 5 06:26:20.005020 as7816-64x ERR snmp#snmp-subagent [sonic_ax_impl] ERROR: No queue stat counters found in the Counter DB. SyncD database is incoherent.",
"Oct 5 06:26:20.006722 as7816-64x ERR snmp#snmp-subagent [ax_interface] ERROR: MIBUpdater.start() caught an unexpected exception during update_data()#012Traceback (most recent call last):#012 File \"/usr/local/lib/python3.6/dist-packages/ax_interface/mib.py\", line 40, in start#012 self.reinit_data()#012 File \"/usr/local/lib/python3.6/dist-packages/sonic_ax_impl/mibs/vendor/cisco/ciscoSwitchQosMIB.py\", line 79, in reinit_data#012 mibs.init_sync_d_queue_tables(self.db_conn)#012 File \"/usr/local/lib/python3.6/dist-packages/sonic_ax_impl/mibs/_init_.py\", line 332, in init_sync_d_queue_tables#012 raise RuntimeError('The queue_stat_map is not defined')#012RuntimeError: The queue_stat_map is not defined",

[jerry-chang3300]

The root cause should be the difference in time between containers.
PortOrch would write QUEUE_COUNTER_ID_LIST to Flex_Counter_DB when initialize, and SyncD_Flex_Counter would write counter to Counter_DB after it collects counter.
But SNMP may query Counter_DB by QUEUE_COUNTER_ID_LIST before SyncD_Flex_Counter write counters to Counter_DB, then SNMP can't get it and write this log.