Shelley monitoring

setup_shelley_monitoring.sh

#!/bin/bash
#---------------------------------------------------------------------
# File:    setup_shelley_monitoring.sh
# Created: 2019/10/17
# Creator: ilap
#=====================================================================
# DESCRIPTION:
#
# This script downloads and configures the required files 
# for monitoring a Shelley node by using grafana/prometheus.
#

clean_up () {
    echo "Cleaning up...." >&2
    rm -rf "${TMP_DIR}"
    RES=$1
    exit ${RES:=127}
}

message() {
    echo -e "$*" >&2
    exit 127
}

get_idx () {
    case $OSTYPE in
        "darwin"*)
            IDX=0
        ;;
        "linux-gnu"*)
            if [[ $HOSTTYPE == *"x86_64"* ]]; then
                IDX=1
            elif [[ $HOSTTYPE == *"arm"* ]]; then
                IDX=2
            else
                message "The $HOSTTYPE  is not supported"
            fi
        ;;
        *)
            message "The \"$OSTYPE\" OS is not supported"
        ;;
    esac
    echo $IDX
}

dl() {
    DL_URL="${1}"
    OUTPUT="${TMP_DIR}/`basename \"${DL_URL}\"`"
    shift
    
    TO_DIR="${@}"
    
    case ${DL} in
        *"wget"*)
        wget --no-check-certificate --output-document="${OUTPUT}" "${DL_URL}";;
        *)
        ( cd ${TMP_DIR} && curl -JOL "${DL_URL}" --silent );;
    esac
}

#### MAIN
if [ -z "$1" ] ; then
    message "usage: `basename $0` <project path> # e.g. ... ./Shelley"
fi

PROJ_DIR="`pwd`/${1}"
PROJ_NAME="`basename ${1}`"

export TMP_DIR=`mktemp -d "/tmp/${PROJ_NAME}.XXXXXXXX"`

CURL=`which curl`
WGET=`which wget`
JCLI=`which jcli`
DL=${CURL:=$WGET}

if  [ -z "$DL" -o -z "`which pip3`" -o -z "$JCLI" ]; then
    message 'You need to have 'wget' or 'curl', 'jcli' and 'pip3' to be installed\nand accessable by PATH environment to continue...\nExiting.'
fi

# Obtain parameters.
IP=127.0.0.1
PORT=3101
export IP PORT

while :
do
    
    read -p "What is the ip of the node (default:${IP})? " ip
    read -p "What is the port of the REST api of the node running on ${IP:="${ip}"}'s (default: ${PORT})? " port
    echo "Is this correct? http://${ip:-"${IP}"}:${port:-"${PORT}"}"
    read -p "Do you want to continue? [Y/n/q] " answer
    
    case ${answer:="Y"} in
        [yY]*)
            IP=${ip:-"${IP}"}
            PORT=${port:-"${PORT}"}
            break;;
        [nN]* )
            continue;;
        [qQ]* )
            exit;;
        * )
            echo "Please enter [yY](es), [nN](o) or [qQ](quit).";;
    esac
done

ARCHS=("darwin-amd64" "linux-amd64"  "linux-armv6")
IDX=`get_idx`

PROM_VER=2.13.0
PROM_URL="https://github.com/prometheus/prometheus/releases/download/v${PROM_VER}/prometheus-${PROM_VER}.${ARCHS[IDX]}.tar.gz"

GRAF_VER=6.4.3
GRAF_URL="https://dl.grafana.com/oss/release/grafana-${GRAF_VER}.${ARCHS[IDX]}.tar.gz"

NIX_URL="https://raw.githubusercontent.com/input-output-hk/jormungandr-nix/master/nixos/jormungandr-monitor"

MON_PY="monitor.py"
GRAF_DB="grafana.json"

trap clean_up  SIGHUP SIGINT SIGQUIT SIGTRAP SIGABRT SIGTERM

echo -e "Downloading prometheus..." >&2
dl $PROM_URL

echo -e "Downloading grafana..." >&2
dl $GRAF_URL

echo -e "Downloading jormungandr monitoring scripts for prometheus" >&2
dl ${NIX_URL}/$MON_PY
dl ${NIX_URL}/$GRAF_DB

cd ${TMP_DIR} && mkdir -p "${PROJ_DIR}"/{exporters,prometheus,grafana}

PROM_DIR="${PROJ_DIR}/prometheus"
GRAF_DIR="${PROJ_DIR}/grafana"
tar zxC "${PROM_DIR}" -f *prome*gz --strip-components 1
tar zxC "${GRAF_DIR}" -f *graf*gz --strip-components 1

echo -e "Configuring components" >&2
cp -pr ${MON_PY} "${PROJ_DIR}"/exporters && chmod +x "${PROJ_DIR}"/exporters/*
cp -pr ${GRAF_DB} "${PROJ_DIR}"/
sed -i -e 's#@jcli@#'"$JCLI"'#g' "${PROJ_DIR}"/exporters/${MON_PY}
sed -i -e 's@\(/usr/bin/env python\)@\13@' "${PROJ_DIR}"/exporters/${MON_PY}
pip3 install ipython python-dateutil prometheus_client >/dev/null

cd ${PROJ_DIR}

EXPORTER_IP=localhost
EXPORTER_PORT=9100
sed -i -e 's@\(^scrape_configs:.*\)@\1\
  - job_name: '\''jormungandr'\''\
    static_configs:\
    - targets: ['\'${EXPORTER_IP}:${EXPORTER_PORT}\'']@g'  "${PROM_DIR}"/prometheus.yml

cat > start_all.sh <<EOF
#!/bin/bash

	#1. exporter
	PORT=${EXPORTER_PORT}
	JORMUNGANDR_API="http://${IP}:${PORT}/api"
	# Addresses to monitor e.g.
	# ADDRESSES="ta1..... ta1..... ta1...."
	ADDRESSES=""
	export PORT JORMUNGANDR_API ADDRESSES
	"${PROJ_DIR}"/exporters/monitor.py &
	sleep 3

	#2. Prometheus
	"${PROM_DIR}"/prometheus --config.file="${PROM_DIR}"/prometheus.yml &
	sleep 3

	#3. Grafana
	#vi conf/defaults.ini
	cd "${GRAF_DIR}"
	./bin/grafana-server web
EOF

chmod a+rx start_all.sh

echo -e "Installation completed

You need to do the following to configure grafana:
0. Start the required services by \"./${PROJ_NAME}/start_all.sh\"
  - Keep in mind this startup script is very simple and has limited capabilities
  - check the monitor script (http://${EXPORTER_IP}:${EXPORTER_PORT})
1. Login to grafana as admin/admin (http://localhost:3000)
2. Add "prometheus" (all lowercase) datasource (http://localhost:9090)
3. Create a new dashboard by importing 'grafana.json' (left plus sign).

Enjoy...
" >&2

clean_up 0

Great Tool... All node HW specific metrics aren't puling any data (CPU, network, Memory) - Jormungnadr metrics working fine... on Mac OSX

I'm using Mac OS Mojave and having problems with the 'ss' command used in monitor.py. Seems Mac doesn't provide "ss"? I see RickyMac68 above said it's working perfectly for him. Have I missed something? I have rudimentary Python skills but don't understand the "@lsof@", "@wc@" and "@ss@ in the monitor.py script. I just replaced the first two with direct paths to the commands and that seemed to fix it for them. But don't have an "ss" command. Any help would be greatly appreciated. Thank you.

Well ... I think I've things working for now. Thanks for the script. It helped a lot.

@Isc3 can you share the changes you made on OSX? thx

I never got 'ss' working. But for now I can get some of these metrics into prometheus and grafana.

I had to pip3 install prometheus_client and py-dateutil python modules.

I basically changed the bash env to python3, commented out everything having to do with "ss". And changed the @lsof@ and @wc@ instances to full paths to each. This just allowed me to run 'python3 monitor.py'. I already have Prometheus/Grafana running. I didn't use the start_all.sh script.
I think that's all of it.
Hopefully this helps.

#!/usr/bin/env python3 <-------- changed from python to python3

JORMUNGANDR_API = os.getenv('JORMUNGANDR_RESTAPI_URL',
os.getenv('JORMUNGANDR_API', 'http://127.0.0.1:<MY_NODE_LISTEN_PORT>/api')) <------- node rest listen port

NODE_METRICS = [
"blockRecvCnt",
"connections",
"lastBlockDate",
"lastBlockEpoch",
"lastBlockFees",
"lastBlockHash",
"lastBlockHeight",
"lastBlockSlot",
"lastBlockSum",
"lastBlockTime",
"lastBlockTx",
"recvq", <------------------ commented out
"txRecvCnt",
"uptime",

Decorate function with metric.
@JORMUNGANDR_METRICS_REQUEST_TIME.time()
def process_jormungandr_metrics():
# Process jcli returned metrics
metrics = jcli_rest(['node', 'stats', 'get'])
lsof = subprocess.Popen(
('/usr/sbin/lsof', '-nPi', ':3000', '-sTCP:ESTABLISHED'), <---------------- Changed @lsof@ to direct path
stdout=subprocess.PIPE)
wc = subprocess.check_output(('/usr/bin/wc', '-l'), stdin=lsof.stdout). <-------------- Changed @wc@ to direct path
lsof.wait()
metrics['connections'] = int(wc, 10)

ss = subprocess.run(('@ss@', '-plntH', '( sport = :3000 )'), stdout=subprocess.PIPE).  <--------------comment out
recvq = ss.stdout.split()[1]                <--------------comment out
metrics['recvq'] = int(recvq, 10)       <--------------comment out

def jcli_rest(args):
flags = ['--host', JORMUNGANDR_API, '--output-format', 'json']
params = ['/usr/local/bin/jcli', 'rest', 'v0'] + args + flags <----------- Put in direct path to my jcli binary
result = subprocess.run(params, stdout=subprocess.PIPE)
return json.loads(result.stdout)