runoncluster

Submits jobs to the tl-app

Installation

pip install runoncluster

Usage

runoncluster script.Rmd inputs.json config.json

Configuration

RMarkdown Script*

Use the params feature for passing inputs to the script:

---
title: "Sample Script"
output: 
  html_document:
    self_contained: false
params:
  sample_size: 10000
---

### Session Information

```{r sessionInfo, echo=FALSE, results="asis"}
sessionInfo()
```

inputs.json

Set the values of parameters as key/value pairs in a json file:

{
	"sample_size": 10000,
	"data": {
		"uri": "https://git.ghap.io/stash/scm/hbgd/ki1000111.git",
		"repository_path": "WASH-BK/adam/full_ki1000111_WASH_BK.csv"
	}		
}

If a data section is included with a git.ghap.io URL, tl-app will clone that repo to the GHAP filesystem and replace the URI with one that points to the file specific in repository_path. This way your RMarkdown script can simply:

dataset = data.table::fread(params$data$uri)

config.json

The configuration file lets the CLI know where to send the job. Here is an example:

{
	"base_url": "https://tl-app-rvit.herokuapp.com/",
	"ghap_username": "${GHAP_USERNAME}",
	"ghap_password": "${GHAP_PASSWORD}",
	"ghap_ip": "${GHAP_IP}",
	"token": "${TLAPP_TOKEN}",
	"r_packages": [
		"knitr",
		"github://jeremyrcoyle/delayed@reduce-r-version",
		"igraph@1.0.1"
	]
}

The special ${VAR_NAME} syntax will read the value from an environment variable before pushing it to the server.

The r_packages section allows you to define R packages to be installed on the target system. There are a few special syntaxes supported:

• <package_name> installs from CRAN
• `<package_name>@ installs a specific version of a package from CRAN
• github://<username>/<package_name> installs from Github using devtools

Right now, you have to ask a developer for an API token. Improved UX coming soon!