Track the RAM usage (resident set size) of a process, its children, its children's children, etc. in real time with a Unicode text sparkline. See the average and the maximum usage after the process exits, as well as the run time.
> memsparkline -- chromium-browser --incognito http://localhost:8081/
▁▁▁▁▄▇▇▇█ 789.5
avg: 371.0
max: 789.5
time: 0:00:12.0
> memsparkline -n -o log du /usr/ >/dev/null 2>&1 &
> tail -f log
█ 2.8
▆█ 3.3
▆▇█ 3.6
▆▇▇█ 3.9
▆▇▇█▆ 3.3
▆▇▇█▆▆ 3.3
▆▇▇█▆▆▆ 3.3
▆▇▇█▆▆▆▆ 3.3
▄▅▅▆▅▅▅▅█ 5.2
▄▅▅▆▅▅▅▅██ 5.2
avg: 3.7
max: 5.2
time: 0:00:10.1
Prebuilt binaries for FreeBSD (amd64), Linux (aarch64, riscv64, x86_64), macOS (arm64, x86_64), OpenBSD (amd64), and Windows (amd64, x86) are attached to releases.
Install Go, then run the following command:
go install github.com/dbohdan/memsparkline@latestmemsparkline works on POSIX systems supported by gopsutil. It has been tested on Debian, Ubuntu, FreeBSD, and OpenBSD. Unfortunately, gopsutil doesn't support NetBSD. NetBSD users can install the last Python release of memsparkline.
Although memsparkline seems to work on Windows, Windows support has received little testing outside of CI. The sparkline displays incorrectly in the Command Prompt and ConEmu on Windows 7 with the stock console fonts. It displays correctly on Windows 10 with the font NSimSun.
Usage: memsparkline [-h] [-v] [-d path] [-l n] [-m fmt] [-n] [-o path] [-q] [-t
fmt] [-w ms] [--] command [arg ...]
Track the RAM usage (resident set size) of a process and its descendants in
real time.
Arguments:
command
Command to run
[arg ...]
Arguments to the command
Options:
-h, --help
Print this help message and exit
-v, --version
Print the version number and exit
-d, --dump path
File to append full memory usage history to when finished
-l, --length n
Sparkline length (default: 20)
-m, --mem-format fmt
Format string for memory amounts (default: '%.1f')
-n, --newlines
Print new sparkline on new line instead of over previous
-o, --output path
Output file to append to ('-' for standard error)
-q, --quiet
Do not print sparklines, only final report
-r, --record ms
How frequently to record/report memory usage in ms (default: 1000)
-s, --sample ms
How frequently to sample memory usage in ms (default: 200)
-t, --time-format fmt
Format string for run time (default: '%d:%02d:%04.1f')
-w, --wait ms
Set '--sample' and '--record' time simultaneously (that both options
override)
memsparkline differentiates between samples and records.
Samples are measurements of memory usage.
Records are information about memory usage printed to the chosen output (given by --output) and added to history (saved using the --dump option).
There is a separate setting for the sample time and the record time. The sample time determines the interval between when memory usage is measured. The record time determines the interval between when a record is made (written to the output and added to history). When sampling is more frequent than recording (as with the default settings), memsparkline uses the highest sampled value since the last record.
A short sample time like 5 ms can result in high CPU usage, up to 100% of one CPU core. To reduce CPU usage, sample less frequently. The default sample time of 200 ms results in memsparkline using around 15% of a 2019 x86-64 core on the developer's machine.
Records are only created when at least one sample has been taken. Setting the record time shorter than the sample time is allowed for convenience, but no record is added when there are no samples.
MIT.
memusg and spark (both linked below) inspired this project.
- DragonFly BSD,
FreeBSD,
macOS,
NetBSD,
and OpenBSD
time(1) flag
-l. - GNU time(1) flag
-v. - memusg — a Bash script for FreeBSD, Linux, and macOS that measures the peak resident set size of a process.
- spark — a Bash script that generates a Unicode text sparkline from a list of numbers.
- sparkline.tcl — a Tcl script inspired by spark made by the developer of this project.
Adds a
--minand--maxoption for setting the scale.