How to calculate cpu utilization of container in docker using http api - docker

Question

I know that the CPU utilization of the container can be obtained by docker stats:
#docker stats
CONTAINER ID NAME CPU % MEM USAGE / LIMIT MEM % NET I/O BLOCK I/O PIDS
05076af468cd mystifying_kepler 0.02% 10.5MiB / 5.712GiB 0.18% 656B / 0B 0B / 0B 1
And I want to get this data through the HTTP api:http api.
Data i get from this http api is :
{
"read": "2019-11-26T22:18:33.027963669Z",
"preread": "2019-11-26T22:18:32.013978454Z",
"pids_stats": {
"current": 1
},
"blkio_stats": {
"io_service_bytes_recursive": [],
"io_serviced_recursive": [],
"io_queue_recursive": [],
"io_service_time_recursive": [],
"io_wait_time_recursive": [],
"io_merged_recursive": [],
"io_time_recursive": [],
"sectors_recursive": []
},
"num_procs": 0,
"storage_stats": {},
"cpu_stats": {
"cpu_usage": {
"total_usage": 361652820,
"percpu_usage": [361652820],
"usage_in_kernelmode": 50000000,
"usage_in_usermode": 100000000
},
"system_cpu_usage": 144599100000000,
"online_cpus": 1,
"throttling_data": {
"periods": 0,
"throttled_periods": 0,
"throttled_time": 0
}
},
"precpu_stats": {
"cpu_usage": {
"total_usage": 361488978,
"percpu_usage": [361488978],
"usage_in_kernelmode": 50000000,
"usage_in_usermode": 100000000
},
"system_cpu_usage": 144598090000000,
"online_cpus": 1,
"throttling_data": {
"periods": 0,
"throttled_periods": 0,
"throttled_time": 0
}
},
"memory_stats": {
"usage": 11005952,
"max_usage": 11108352,
"stats": {
"active_anon": 11005952,
"active_file": 0,
"cache": 0,
"dirty": 0,
"hierarchical_memory_limit": 9223372036854771712,
"hierarchical_memsw_limit": 9223372036854771712,
"inactive_anon": 0,
"inactive_file": 0,
"mapped_file": 0,
"pgfault": 8151,
"pgmajfault": 0,
"pgpgin": 4137,
"pgpgout": 1450,
"rss": 11005952,
"rss_huge": 0,
"total_active_anon": 11005952,
"total_active_file": 0,
"total_cache": 0,
"total_dirty": 0,
"total_inactive_anon": 0,
"total_inactive_file": 0,
"total_mapped_file": 0,
"total_pgfault": 8151,
"total_pgmajfault": 0,
"total_pgpgin": 4137,
"total_pgpgout": 1450,
"total_rss": 11005952,
"total_rss_huge": 0,
"total_unevictable": 0,
"total_writeback": 0,
"unevictable": 0,
"writeback": 0
},
"limit": 6133108736
},
"name": "/mystifying_kepler",
"id": "05076af468cdeb3d15d147a25e8ccee5f4d029ffcba1d60f14f84e2c9e25d6a9",
"networks": {
"eth0": {
"rx_bytes": 656,
"rx_packets": 8,
"rx_errors": 0,
"rx_dropped": 0,
"tx_bytes": 0,
"tx_packets": 0,
"tx_errors": 0,
"tx_dropped": 0
}
}
}
I was able to calculate the utilization of memory from the data, but I didn't find a way to get the CPU utilization .
And ideas?

Answer 1

You've probably solved this by now, but for the next person... This example is in Python, but the data fields and math are the same if you're making API calls.
The API returns cumulative values, so you need more than one sample - do the math using the difference between samples to get the utilization for that period. This example uses the streaming mode, which pushes an update every second.
# These initial values will seed the "last" cycle's saved values
containerCPU = 0
systemCPU = 0
container = client.containers.get(containerID)
#This function is blocking; the loop will proceed when there's a new update to iterate
for stats in container.stats(decode=True):
#Save the values from the last sample
lastContainerCPU = containerCPU
lastSystemCPU = systemCPU
#Get the container's usage, the total system capacity, and the number of CPUs
#The math returns a Linux-style %util, where 100.0 = 1 CPU core fully used
containerCPU = stats.get('cpu_stats',{}).get('cpu_usage',{}).get('total_usage')
systemCPU = stats.get('cpu_stats',{}).get('system_cpu_usage')
numCPU = len(stats.get('cpu_stats',{}).get('cpu_usage',{}).get('percpu_usage',0))
# Skip the first sample (result will be wrong because the saved values are 0)
if lastContainerCPU and lastSystemCPU:
cpuUtil = (containerCPU - lastContainerCPU) / (systemCPU - lastSystemCPU)
cpuUtil = cpuUtil * numCPU * 100
print(cpuUtil)