llnl-sina 1.14.0

Simulation INsight and Analysis

Contents

• Overview

• LC Setup

• Standard Setup

• Manual Setup

  • Creating the Environment

  • Installing Software Dependencies

• Activating and Deactivating the Virtual Environment

• Using the Example Notebooks and Datasets

• Testing

• Supported Environments

• Database Support

Overview

Sina’s Python component is a tool for making simulation (meta)data collection and exploration simple.

It works by collecting information from code runs, logs, and other outputs into a common file format which can then be passed off to one of Sina’s supported backends, all of which are queried using the same user-friendly Python API. To the end user, this means that important data can be accessed through Python scripts, GUIs (Jupyter notebooks) etc. with all the speed of a database and none of the complexity (the user never has to interact with database architecture), nor any of the traditional headaches of parsing logs or remembering which file contains what.

Sina is integrated into a number of LLNL physics codes to capture simulation data; look for the _sina.json! If your code isn’t configured to output Sina, but you’d like it to be, we may be able to work with the code team to integrate it–you can reach us at [e-mail], or check the WEAVE project.

The instructions below will guide you through setting up a virtual environment for Sina (or installing it in one that already exists), running example notebooks, and getting dependencies for your backend(s) of choice. Note that SQL will always be available as the “default” backend. Once you’re done with setup, a quickstart tutorial can be found in notebook form at <sina_root>/examples/basic_usage.ipynb.

Remember that, if you’re on LC, each time you log in you’ll first need to activate the environment. When you’re done, we recommend you deactivate the virtual environment to get back to your default environment or end your session.

LC Setup

If you’re on an LC machine, you can use a virtual environment with dependencies already installed:

$ source /collab/usr/gapps/wf/releases/sina/bin/activate

The above is for bash; other activation scripts, e.g. activate.csh, can be found in the same directory.

Sina will now be available for use via Python virtual environment, and can be tested with sina -h (which should display a help message). When you’re done, use deactivate to exit the virtual environment. Note that this will be the release (master) Sina version–if you want to use Sina Develop, keep reading!

If you run into issues with the LC virtual environment, please email us at [e-mail].

Standard Non-LC Setup

Sina is available on PyPi:

$ pip install llnl-sina

However, this will only give you access to the release version! Non-release versions are not available externally. Internal users looking to use our development version, or wanting to contribute to Sina, clone us from CZ Gitlab. External contributors should clone us from the LLNL Github.

After cloning, cd to the python folder. Standard installation, provided by the Makefile, is initiated by entering the following at the command line:

$ make

This command creates the virtual environment, installs (missing) dependencies, and installs Sina.

You can build the documentation, which will appear in build/docs, using:

$ make docs

Tests are run by entering:

$ make tests

Alternatively, you can run all of the above by entering the following:

$ make all

This will install Sina with its default backend (SQL).

Once installation is complete, you are ready to activate the environment – see “Activating the Virtual Environment” – and begin using Sina. You can also install the optional Cassandra backend with:

$ make cassandra

Additional command line tools (such as diffing Records) are available with:

$ make cli-tools

Manual Setup

You will need to create, activate, and install software dependencies in your environment.

Creating the Environment

Enter the following command to set up your initial environment:

$ python -m virtualenv venv   # venv can be any name you want

Now activate the environment (see “Activating the Virtual Environment”).

You need to be in the proper Sina directory before proceeding to install dependencies, so enter the following command:

(venv) $ cd $SINA_PYTHON

where SINA_PYTHON is the python subdirectory of the Sina source code.

Installing Software Dependencies

You first need to make sure there is a requirements/links.txt file that contains the appropriate link constraints. There are two requirements files containing flags and links used in our supported environments:

- requirements/lc-links.txt
- requirements/no-links.txt

The first file contains the options needed to restrict packages to those available on the wheelhouse hosted on the Open Computing Facility (OCF) at Lawrence Livermore National Laboratory. The second file contains no flags. The links.txt file is included in other requirements files to ensure the options are consistent for the build and testing processes. The makefile will create a softlink to the appropriate file; if you’re doing this manually, you’ll need to link requirements/links.txt to the appropriate file yourself.

Once you’ve set up your requirements/links.txt, you can use our dev requirements file (<sina_root>/python/requirements/development.txt) to install basic Sina dependencies:

$(venv) pip install -r requirements/development.txt

The requirements file should install the package in editable mode but, if not, run:

$(venv) pip install -e .

Activating and Deactivating the Virtual Environment

Enter the following command to enter the virtual environment:

$ source $SINA_PYTHON/venv/bin/activate  # use activate.csh if in a [t]csh

where SINA_PYTHON is the python subdirectory of the Sina source code. You will need to do this every time you want to start up a session in the named virtual environment. You can test Sina’s available with:

(venv) $ sina --version

Enter the following command to deactivate the virtual environment:

(venv) $ deactivate

when you are done.

Using the Example Notebooks and Datasets

Sina contains tutorials in the form of Jupyter notebooks. Files are stored in the examples directory (found in the sina root folder alongside the python and cpp folders), and are organized by dataset, with data_overview.rst containing descriptions of each set. To use the notebooks, you’ll first need to run getting_started.ipynb (also in the examples directory) from the LC Jupyter server at lc.llnl.gov/jupyter. This will create a Jupyter kernel from your current virtual environment, making anything installed in it available to the notebook. After that, you’ll be ready to run the rest of the notebooks. If you’re not working on LC, you can also set Jupyter up locally: run make Jupyter from the python folder, then jupyter notebook. This will open a webpage similar to what you’d see accessing LC’s Jupyter server.

Most notebooks rely on sample datasets. Pre-built sets are deployed with Sina to the LC, but you can build them locally as well to experiment with Sina. Go into any dataset folder (the NOAA set is well-sized for experimentation) and ./build_db.sh. Note that you’ll need Sina available to do so, see the section on virtual environments.

To clean all output from the notebooks:

(venv) $ make clean-notebooks

Testing

This package uses pytest to run unit tests. Enter the following while in your virtual environment:

(venv) $ pytest

Additional tests, which include checks for PEP8 compliance and proper documentation, can be run my entering the following:

$ make tests

This command will set up and enter the necessary virtual environment.

Supported Environments

Sina is regularly tested in the following environments:

• OSX 10.15: Primary development environment for most team members. If you are not on the LC network, be sure to comment out –no-index in the requirements file.

• TOSS 3, RedHat 7.4 (quartz, rzsonar): Automated testing environment

• TOSS 3, RedHat 7.5 (catalyst, rztopaz): Secondary development environment

Absence is not an indication that Sina will not work; please consider expanding this list!

Database Support

Out-of-the-box, Sina does not install drivers for relational databases other than SQLite. If you wish to connect to other databases (e.g. MySQL, MariaDB, or Oracle), you need to install the appropriate drivers for that database. You can do this with our Makefile:

$ make mysql

After you install the connector, you can connect to these types of databases from the command line tools:

$ sina ingest --database-type=sql --database "mysql+mysqlconnector://host:port/?read_default_file=~/.my.cnf"

You can also connect with the programmatic API:

factory = DAOFactory("mysql+mysqlconnector://host:port/?read_default_file=~/.my.cnf")