pagurus

Used for getting the information from a running job based on its pid using psutil.

Install

Availible on PyPi and easily installed in your enviroment.

pip install pagurus

Options

usage: pagurus [-h] [-o OUTFILE] [-p PATH] [-d] [-r RATE] [-u USER] [-noh] [-mv] [-l ROLLING] [--json] [--envvar ENVVAR]

options:
  -h, --help            show this help message and exit
  -o OUTFILE, --outfile OUTFILE
                        File name for csv.
  -p PATH, --path PATH  Path to put csv file.
  -d, --debug           Run with debugging info.
  -r RATE, --rate RATE  Polling rate for process.
  -u USER, --user USER  Username to get stats for.
  -noh, --no-header     Turn off writting the header.
  -mv, --move           Moves file from 'running' to 'done' directories
  -l ROLLING, --rolling ROLLING
                        Time to roll file over to number to file name in ~minutes.
  --json                Output JSON strings instead of CSV lines
  --envvar ENVVAR       add environment var to output (can be specified multiple times)

Running pagurus as a wrapper for a single user

# Start running wrapper in the background for username
pagurus -u $USER -mv -p /path/to/output/dir -o test.csv
# Get the previous running PID of pagurus
export PID=$!
# Sleep for a few seconds to let everything start running
sleep 10 

###########################

# Run your desired program as normal
./a.out

# Works with containers
shifter --image=tylern4/memoryhog:latest alloc 2

# and with wrapper scripts
shifter --image=jfroula/aligner-bbmap:2.0.2 bbmap.sh Xmx12g in=sample.fastq.bz2 ref=sample.fasta out=test.sam

###########################

# Kill the pagurus process
kill $PID

# Sleep for a few seconds to let results finish writing
sleep 10

Plotting results example

There is an example notebook which will shows how to get memory usage and cpu usage from the output files.