sg-collectinfo

README.md

Collecting

1. Clone this gist
2. chmod *.sh
3. ./collect.sh

Sending

1. zip up the output directory
2. Send/upload zipfile
3. delete the output directory

collect.sh

mkdir -p output

while true; do

    # Collect netstat output
    # This must be run as sudo, because otherwise it will give errors
    # for the "netstat -p" about not being root.  (even though running
    # netstat -p on the command line works fine)
    sudo python netstat.py >> output/netstat.txt

    # Collect top output
    ./top.sh >> output/top.txt

    # Collect expvars output
    curl http://localhost:4985/_expvar >> output/expvars.txt

    sleep 60

done

netstat.py

import subprocess
import json

# Collect the number of connection types from
# the netstat output for the sync_gateway process.
# The result will be a json file like:
#
# {"ESTABLISHED": 250, "TIME_WAIT": 100}
#
def get_filtered_netstat():
    output = get_raw_netstat_output()
    lines = filter_netstat_header(output)
    lines = filter_non_sync_gateway(lines)
    stats = generate_stats(lines)
    emit_stats_json(stats)

def filter_netstat_header(output):
    # convert from string to array of strings
    lines = output.split("\n")
    return lines[2:]  # slice off the first two lines, return the rest

def filter_non_sync_gateway(lines):
    # tcp        0      0 10.181.112.50:34656     10.157.88.10:11210      ESTABLISHED 1001       121024     6925/sync_gateway
    remaining = []
    for line in lines:
        if "sync_gateway" in line:
            remaining.append(line)
    return remaining

def get_raw_netstat_output():
    # Proto Recv-Q Send-Q Local Address           Foreign Address         State       User       Inode      PID/Program name
    # tcp        0      0 10.181.112.50:34656     10.157.88.10:11210      ESTABLISHED 1001       121024     6925/sync_gateway
    # tcp        0      0 10.181.112.50:38528     10.109.187.75:11210     ESTABLISHED 1001       121039     6925/sync_gateway
    return subprocess.check_output(["netstat", "-apeen"])

def generate_stats(lines):

    # key: state (eg, ESTABLISHED)
    # val: counter (num times this has been seen)
    stats = {}

    for line in lines:
        # parse the state field
        columns = line.split()
        state = columns[5]
        if not stats.has_key(state):
            stats[state] = 1
        else:
            current_count = stats[state]
            stats[state] = current_count + 1

    return stats

def emit_stats_json(stats):
    print json.dumps(stats)

if __name__ == "__main__":
    get_filtered_netstat()

top.sh

# - Call top
# - Grep the line with sync_gateway
# - Parse out these fields:
#   VIRT
#   RES
#   %CPU
# - Output JSON: {"VSS":1976804 "RSS":1.291g "CPU":0.0}

top -n 1 -b | grep -i sync_gateway | awk '{print "{\"VSS\":" $5 " \"RSS\":" $6 " \"CPU\":" $9 "}"}'