KProbe

Name	Functionality
cpudist	This kprobe measures the time a task spends on or off the CPU, and shows this time s a histogram, optionally per-process.
pidpersec	This kprobe shows the number of new processes created per second, measured by tracing
runqlat	This kprobe summarizes scheduler run queue latency as a histogram, showing

# run the cpu_kprobe
sudo python3 ./kprobes/jx_cpu_kprobe.py -d kprobes/db/metrics.json

# run the pidpersec_kprobe
sudo python3 ./kprobes/pidpersec.py -d kprobes/db/metrics.json

# run the runqlat_kprobe
sudo python3 ./kprobes/runqlat.py -d kprobes/db/metrics.json

some fixed arguments for cpu_kprobe

args.extension = 1
args.pid = 3327
args.interval = 1
# must use extension, interval is 1s, and pid is fixed as 3327

Preprocessing Functions

function name
cpudist_avg
cpudist_sum
pidpersec_avg
pidpersec_sum
runqlat_avg
runqlat_sum

V0

run a KProbe

sudo python3 ./cpudist.py -p 1888 -e 1

V1

worker

• the worker will run a Flask server in its main function and run a user agent in another thread (attach a kprobe, extract information and store it in db)
• each worker has 1 specific port number and 1 database to store the extracted metrics

make runWorker1
make runWorker2
make runWorker3

client

# run the client
make runClient

# input a command
select cpu_avg,cpu_sum from n1,n2,n3
# metrics_name will be "cpu_avg,cpu_sum"
# nodes will be "n1,n2,n3"

select cpu_avg from n1,n2,n3

select cpu_avg,cpu_sum from n1,n3

select cpu_avg,pidpersec_avg from n1,n2,n3

select cpu_avg,pidpersec_avg,runqlat_avg,runqlat_sum from n1,n2,n3

V2

worker

make runWorker1
make runWorker2
make runWorker3

coordinator

make runCoordinator

client

# run the client
make runClient


# operations
select cpu_avg from n1,n2,n3

select cpu_avg,cpu_sum from n1,n3

select cpu_avg,pidpersec_avg from n1,n2,n3

select cpu_avg,pidpersec_avg,runqlat_avg,runqlat_sum from n1,n2,n3