A handy all-in-one raintank/graphite-stack (from Grafana's authors) image and psutil and statsd client. It may seem an overkill for a simple one-off test, but for something like a several-day debugging it's, for sure, reasonable. pip install memory_profilerīy default this pops up a Tkinter-based ( python-tk may be needed) chart explorer which can be exported: It can also record process with its children processes (see mprof -help). The package provides RSS-only sampling (plus some Python-specific options). Psrecord $(pgrep proc-name2) -interval 1 -duration 60 -plot plot2.png & Psrecord $(pgrep proc-name1) -interval 1 -duration 60 -plot plot1.png & Sudo apt-get install python-matplotlib python-tk # for plotting or via pipįor single process it's the following (stopped by Ctrl+C): psrecord $(pgrep proc-name1) -interval 1 -plot plot1.pngįor several processes the following script is helpful to synchronise the charts: #!/bin/bash
pip install psrecord # local user install Python psrecord package does exactly this. The following addresses history graph of some sort. Procpath plot -d ff.sqlite -q rss -p 123 -f rss.svgĬharts look like this (they are actually interactive Pygal SVGs): procpath record -i 1 -r 120 -d ff.sqlite '$.children[?("firefox" in RSS and CPU usage of a single process (or several) out of all recorded would look like: procpath plot -d ff.sqlite -q cpu -p 123 -f cpu.svg This records all processes with "firefox" in their cmdline (query by a PID would look like = 42)]') 120 times one time per second. It's a pure-Python CLI package including its couple of dependencies (no heavy Matplotlib), can potentially plot many metrics from procfs, JSONPath queries to the process tree, has basic decimation/aggregation (Ramer-Douglas-Peucker and moving average), filtering by time ranges and PIDs, and a couple of other things.
Returning to the problem of process analysis frequently enough and not being satisfied with the solutions I described below originally, I decided to write my own.
0 kommentar(er)
