Per-CPU usage is unavailable on some CPUs

[gridhead]

This is a snapshot from the frontend scrapping data from the driver running on an Intel CPU.

And this is the output from the container statistics endpoint

While the following is the snapshot from the frontend scrapping data from the driver running on a Broadcom CPU.

And this is the output from the container statistics endpoint

The reason behind such an occurrence is unknown but this thing breaks the statistics page so this should be fixed promptly.

[gridhead]

This needs to be fixed at the frontend side as the driver emits the data that has been made available by the Docker Py module.

Python 3.9.2 (default, Feb 20 2021, 18:40:11) 
Type 'copyright', 'credits' or 'license' for more information
IPython 7.22.0 -- An enhanced Interactive Python. Type '?' for help.

In [1]: from docker import DockerClient

In [2]: dock = DockerClient()

In [7]: dock.containers.list()
Out[7]: [<Container: a940a858b8>]

In [8]: a = dock.containers.list()[0]

In [9]: a.stats
Out[9]: <bound method Container.stats of <Container: a940a858b8>>

In [10]: a.stats()
Out[10]: <generator object APIClient._stream_helper at 0x7f54c0299270>

In [11]: a.stats(stream=False)
Out[11]: 
{'read': '2021-04-03T01:43:51.278450805Z',
 'preread': '2021-04-03T01:43:50.276025491Z',
 'pids_stats': {'current': 1, 'limit': 9366},
 'blkio_stats': {'io_service_bytes_recursive': [{'major': 8,
    'minor': 0,
    'op': 'read',
    'value': 3973120},
   {'major': 8, 'minor': 0, 'op': 'write', 'value': 0}],
  'io_serviced_recursive': None,
  'io_queue_recursive': None,
  'io_service_time_recursive': None,
  'io_wait_time_recursive': None,
  'io_merged_recursive': None,
  'io_time_recursive': None,
  'sectors_recursive': None},
 'num_procs': 0,
 'storage_stats': {},
 'cpu_stats': {'cpu_usage': {'total_usage': 35105000,
   'usage_in_kernelmode': 27003000,
   'usage_in_usermode': 8101000},
  'system_cpu_usage': 1983850000000,
  'online_cpus': 4,
  'throttling_data': {'periods': 0,
   'throttled_periods': 0,
   'throttled_time': 0}},
 'precpu_stats': {'cpu_usage': {'total_usage': 35105000,
   'usage_in_kernelmode': 27003000,
   'usage_in_usermode': 8101000},
  'system_cpu_usage': 1979830000000,
  'online_cpus': 4,
  'throttling_data': {'periods': 0,
   'throttled_periods': 0,
   'throttled_time': 0}},
 'memory_stats': {'usage': 4902912,
  'stats': {'active_anon': 0,
   'active_file': 2027520,
   'anon': 540672,
   'anon_thp': 0,
   'file': 3784704,
   'file_dirty': 0,
   'file_mapped': 2838528,
   'file_writeback': 0,
   'inactive_anon': 540672,
   'inactive_file': 1757184,
   'kernel_stack': 49152,
   'pgactivate': 396,
   'pgdeactivate': 0,
   'pgfault': 1716,
   'pglazyfree': 0,
   'pglazyfreed': 0,
   'pgmajfault': 0,
   'pgrefill': 0,
   'pgscan': 0,
   'pgsteal': 0,
   'shmem': 0,
   'slab': 0,
   'slab_reclaimable': 0,
   'slab_unreclaimable': 0,
   'sock': 0,
   'thp_collapse_alloc': 0,
   'thp_fault_alloc': 0,
   'unevictable': 0,
   'workingset_activate': 0,
   'workingset_nodereclaim': 0,
   'workingset_refault': 0},
  'limit': 8199704576},
 'name': '/trusting_clarke',
 'id': 'a940a858b80c6795051eedec4f7397830290b74b63216de160b00ddd732e0058',
 'networks': {'eth0': {'rx_bytes': 4832,
   'rx_packets': 40,
   'rx_errors': 0,
   'rx_dropped': 0,
   'tx_bytes': 0,
   'tx_packets': 0,
   'tx_errors': 0,
   'tx_dropped': 0}}}

Notice how Docker Py does not provide the per-CPU usage information in the output of the a.stats(stream=False) command.

[gridhead]

Fixed at the frontend side with gridhead/supervisor-frontend-service#85. No action is required for the driver side of things.