Plotting resource usage statistics

Using monitor to generate CPU and memory use statistics for a process and its children:

monitor -i 0.2 -o /tmp/usage.tsv 'ebk-find -n'

Here I am using ebk-find with fzf as the selection command.

The Monitor output looks like this:

Read the data into Python using Pandas:

import pandas as pd
data = pd.read_table("/tmp/usage.tsv")
grouped = data.groupby("Process")

Set up a figure for Matplotlib to plot to, with two subplots: one for CPU usage and one for memory:

import matplotlib.pyplot as plt
plt.style.use("bmh")
fig = plt.figure(figsize=(6, 6), layout="constrained")
mem, cpu = fig.subplots(nrows=2, sharex=True)

For each process in the table, plot the CPU usage against time:

for cmd, subset in grouped:
    cpu.plot("Time", "CPU%", "o-", data=subset, label=cmd)

cpu.set_xlabel("Time, s")
cpu.set_ylabel("CPU, %")

Plot the memory usage (RSS or VMS) against time in the other subplot. Since we have a shared x-axis, we don’t set an x label this time.

for cmd, subset in grouped:
    mem.plot("Time", "RSS", "o-", data=subset, label=cmd)

mem.set_ylabel("RSS memory, MiB")

Add the legend to one of the subplots (whichever has the most white space):

cpu.legend()

Then save the figure to file:

fig.savefig("/tmp/usage.svg", transparent=True)