containered-parflowvr

ParFlowVR in containers!

This project is used to install parflow with FlowVR extension (parFlowVR) comfortably into a docker environment

WARNING: the docker containers in this repository disable some major security features to simplify (remote) debugger access. Thus DO NOT EXPOSE THEM DIRECTLY TO THE INTERNET or to other untrusted networks.

HowTo:

tested on ubuntu 16

install docker

sudo apt-get install docker.io
sudo groupadd docker
sudo usermod -aG docker $USER

now log out and in again and check that you are in the docker group (id) Also check that your docker API version is at least 1.25 (docker version).

clone this repository and run the build script. The build script will take some time (up to hours depending on your internet connection and hard drive speed)

git clone https://github.com/xy124/containered-parflowvr.git
cd containered-parflowvr
./build.sh fast

run the built image to create the container (only needed to be done once after build)

./run.sh

connect into the container

./connect.sh

All data stored to folders other than /home/docker in the container will be deleted if the container is deleted.
The dockerhome/ dir is common between the container and the host machine. In the container it is mounted as /home/docker

you should now be in the docker container having a shell prompt that looks like this:

docker@410aef2ef73e:~$

install the last things in the docker container (netcdf-c, flowvr-ex, parflow)

cd /home/docker/recipes
./build-from-recipes.sh

Before running the testcases you need to log out and in again so that the just installed envrionment scripts for flowvr will be loaded.

Here you are! Play the testcases:

In Situ writing to get better performances

cd /home/docker/parflow/flowvr/testcases/netcdfwriter
./do.sh

In Situ steering of a running Simulation

This shows how to interact live and online with a running parflow simulation

cd /home/docker/parflow/flowvr/testcases/steering-Python
./do.sh

In Situ visualization

Further you can kick off VisIt which was installed to the container to perform In Situ visualization of running parflow simulations

visit &
cd /home/docker/parflow/flowvr/testcases/visit
./do.sh

now bring the visit window to the foreground
click on File -> Open
navigate to /home/docker/.visit/simulations
doubleclick the .sim2 file
click Add -> Pseudocolor and select the variable you desire to inspect
click Draw and here you go
click File -> Simulations . . .
click trigger snap and wait a moment... the plot will refresh.

Visualize a parflowvr application's data flow graph

Change to the applications directory for which you want to spectate the graph and open parflowvr.net.xml with flowvr-glgraph, e.g.

cd /home/docker/parflow/flowvr/testcases/visit
flowvr-glgraph parflowvr.net.xml

(If there is no such parflowvr.net.xml file just execute the app with ./do.sh, this will generate it)

HINT: if VisIt does not show up correctly this is probably related to your graphics drivers that are not exposed correctly to docker. You might want to change the run.sh script accordingly in this case (see http://gernotklingler.com/blog/howto-get-hardware-accelerated-opengl-support-docker/). To re-run your parflowvr-run container

./stop.sh
docker rm parflowvr-run
./run.sh

Bonus: Start/Stop the container (on the host)

./start.sh
./stop.sh

Compile and install own versions of parflow / flowvr-ex

we assume you are on the docker container
parflow code is in /home/docker/parflow
flowvr-ex code is in /home/docker/flowvr-ex
to build:
- go to the code dir (/home/docker/{parflow|flowvr-ex})
- go into the build dir
- optional: if you want to reconfigure with cmake: ccmake ..
  - remember to keep the CMAKE_INSTALL_PREFIX=/home/docker/install intact!
- to compile and install: make install -j$N

Troubleshooting

./stop.sh, ./start.sh, ./run.sh
sudo systemctl restart docker
killall dockerd
docker rm parflowvr-run
Installing docker on debian can be more complicated. Sometimes manually downloading the docker.deb files helps.

build - run - start - stop

build.sh builds the docker container image without running it
run.sh initializes a docker container and starts it
stop.sh stops a docker container
start.sh starts a stopped docker container

In general it is enough to call build.sh and run.sh once. If something went wrong or you just want to stop the docker container to have e.g. more RAM available execute ./stop.sh To start it later again: use ./start.sh

Container structure

An image (parflow-base) is created on build containing ubuntu with all the necessary dependencies installable through the package manager
On top of this an other Dockerfile creates the image parflowvr-run which then is instantiated in the run.sh script.
This allows to change settings and other stuff without reinstalling all the packages again.

Further Parflow containers

This repository also specifies containers to run parflow in a vanilla state (using the official parflow version without many additionally features) (see parflow-vanilla/README.md) plus there is also a chocolate version which consists of the official parflow version with netCDF activated.

Further information on ParFlowVR

The following resources are useful to understand FlowVR and how it was integrated with Parflow:

documentation on http://flowvr.sourceforge.net/
main.pdf on https://gitlab.cs.fau.de/friese/MasterSupplement
the source on https://github.com/xy124/parflow/tree/parFlowVR

xy124 / containered-parflowvr Goto Github PK