grafana

---

Instructions on How To Run PCM Grafana Dashboard

Installation on target system to be analyzed:

1. Build from the root PCM directory:

   cd ../..  # if the current location is 'scripts/grafana/'
   mkdir build && cd build
   cmake .. && make -j$(nproc) pcm-sensor-server
   

2. As root start pcm-sensor-server: cd bin && sudo ./pcm-sensor-server

Alternatively one can start pcm-sensor-server as a container from docker hub.

Installation of the grafana front-end (can be on any host system with connectivity to the target system):

1. Make sure curl and docker are installed on the host
2. In PCM source directory on the host: cd scripts/grafana
3. (Download once and) start docker containers on the host: sudo bash start.sh http://target_system_address:9738
   • start.sh script starts telegraf/influxdb/grafana containers
   • start-prometheus.sh is an alternative script which starts prometheus + grafana containers: sudo bash start-prometheus.sh target_system_address:9738
   • start.sh and start-prometheus.sh can also be used to monitor multiple hosts running pcm-sensor-server containers:sudo bash start.sh targets.txt or sudo bash start-prometheus.sh targets.txt. Here targets.txt should be of the following format:

   host1_ipaddress:pcmport
   host2_ipaddress:pcmport
   .
   .
   hostn_ipaddress:pcmport

   • Don't use localhost to specify the target_system_address if the host and the target are the same machine because localhost resolves to the own private IP address of the docker container when accessed inside the container. The external IP address or hostname should be used instead.
4. Start your browser at http://host:3000/ and then login with admin user, password admin . Change the password and then click on "Home" (left top corner) -> "Dashboards" -> "Intel® Performance Counter Monitor (Intel® PCM) Dashboard"
5. You can also stop and delete the containers when needed: sudo bash stop.sh