TensorBoard + ngrok tunnel: access to tf logs from anywhere

ngrok.yml

authtoken: <copy key from https://dashboard.ngrok.com/auth>
log_level: error
log: stderr

run_tb.sh

#!/bin/bash

# path to tensorboard logs
LOGS_DIR="checkpoint/run1"


# ngrok.yml need contains `log_level` and `log`, because api not working without it
# https://github.com/inconshreveable/ngrok/issues/313


ROOT=`pwd`
RED='\033[0;31m'
GREEN='\033[0;32m'
NC='\033[0m' # No Color


if [ ! -f "${ROOT}/ngrok-stable-linux-amd64.zip" ]
then
    echo 'download ngrok...'
    wget https://bin.equinox.io/c/4VmDzA7iaHb/ngrok-stable-linux-amd64.zip
fi


if [ ! -f "${ROOT}/ngrok" ]
then
    echo 'unzip ngrok...'
    unzip ngrok-stable-linux-amd64.zip
fi


CHECK_TB=`ps | grep tensorboar`
if [ -z "${CHECK_TB}" ]
then
    echo 'run tensorboard...'
    tensorboard --logdir ${LOGS_DIR} --host 0.0.0.0 --port 6006 > /dev/null 2>&1 &
fi


CHECK_NGROK=`ps | grep ngrok`
if [ -z "${CHECK_NGROK}" ]
then
    ./ngrok http -config=ngrok.yml 6006 > /dev/null 2>&1 &

    for i in {1..10}
    do
        sleep 1
        CHECK_NGROK=`ps | grep ngrok`
        if [ ! -z "${CHECK_NGROK}" ]
        then
            sleep 1 # fix 'IndexError: list index out of range'
            break
        fi
    done
fi


echo -e "${GREEN}Ngrok tunnel:${NC}"
curl -s http://127.0.0.1:4040/api/tunnels | python3 -c \
    "import sys, json; print(json.load(sys.stdin)['tunnels'][0]['public_url'])"