Writing own Dockerfile example

13-docker/13-docker.Rmd

@@ -61,7 +61,7 @@ knit_hooks$set(
   - [Base R](#r-base)
   - [RStudio+](#rstudio+)
 
-4. [Building your own Dockerfiles & images](#building)
+4. [Writing your own Dockerfiles & images](#writing)
 
 5. [Sharing files with a container](#share)
 
@@ -248,9 +248,9 @@ name: examples
 
 It should now be clear that Docker is targeted at (and used by) a bewildering array of software applications.
 
-In the realm of economics and data science, that includes every major open-source programming language and software stack.<sup>1</sup> For example, you could download and run a [Julia container](https://hub.docker.com/_/julia/) right now if you so wished.
+In the realm of economics and data science, that includes every major open-source programming language and software stack.<sup>†</sup> For example, you could download and run a [Julia container](https://hub.docker.com/_/julia/) right now if you so wished.
 
-.footnote[<sup>1</sup> It's possible to build a Docker image on top of proprietary software ([example](https://github.com/mathworks-ref-arch/matlab-dockerfile)). But license restrictions make this complicated. I've rarely seen it done in practice.]
+.footnote[<sup>†</sup> It's possible to build a Docker image on top of proprietary software ([example](https://github.com/mathworks-ref-arch/matlab-dockerfile)). But license restrictions make this complicated. I've rarely seen it done in practice.]
 
 --
 
@@ -286,7 +286,7 @@ A quick note on these `docker run` flags:
 
 # Example 1: Base R (cont.)
 
-As promised, here is a GIF of me running the command on my system. The whole thing takes about a minute and takes me directly into an R session.
+As promised, here is a GIF of me running the command on my system. The whole thing takes about a minute and launches directly into an R session.
 
 ```{r docker_r_base_gif, eval = TRUE, echo=FALSE, out.width="75%"}
 knitr::include_graphics("pics/docker-r-base.gif")
@@ -372,6 +372,7 @@ Let's try the [`tidyverse`](https://hub.docker.com/r/rocker/tidyverse) image fro
 *Again, this next line will take a minute or three to download and extract the first time. But the container will be ready for immediate deployment on your system thereafter.*
 
 ---
+name: tverseinit
 count: false
 
 # Example 2: RStudio+ (cont.)
@@ -468,26 +469,27 @@ I can also load the **tidyverse** straight away. (We can ignore those warning me
 
 # Example 2: RStudio+ (cont.)
 
-To stop this container, open up a new terminal window. Grab the container ID if you've forgotten it with `$ docker ps`. Then run:
+To stop this container, you would grab the container ID (i.e. with `$ docker ps`) and then run:
 
 ```{bash}
 docker stop <containerid>
 ```
 
+Please don't do this yet, however! I want to continue using this running container in the next section.
+
 --
 
-</b>
+</b></br>
 
 **Aside:** Recall that we instantiated this container as a detached/background process (`-d`).
 
 ```{bash}
 docker run `-d` -p 8787:8787 -e PASSWORD=pswd123 rocker/tidyverse:4.0.0
 ```
 
-If you dropped the `-d` flag and re-ran the above command, your terminal would stay open as an ongoing process. (Try this yourself.)
-- Everything else would remain the same. You'd still navigate to `<IPADDRESS>:8787` to log in, etc.
-- However, I wanted to mention this non-background process version because it offers another way to shut down the container: Simply type `CTRL+c` in the (same, ongoing process) Terminal window. Again, try this yourself.
-- Confirm that the container is stopped by running `$ docker ps`.
+If you dropped the `-d` flag and re-ran the above command, your terminal would stay open as an ongoing process. (Try this for yourself later.)
+- Everything else would stay the same. You'd still log in at `<IPADDRESS>:8787`, etc.
+- However, I wanted to mention this non-background process version because it offers another way to shut down the container: Simply type `CTRL+c` in the (same, ongoing process) Terminal window. Again, try this for yourself later.
 
 ---
 
@@ -525,37 +527,43 @@ All of which provides a nice segue to our next section...
 
 ---
 class: inverse, center, middle
-name: building
+name: writing
 
-# Building your own Dockerfiles & images
+# Writing your own Dockerfiles & images
 <html><div style='float:left'></div><hr color='#EB811B' size=1px width=796px></html>
 
 ---
 
 # Add to an existing container
 
-The easiest way to start building our own Docker images is by layering on top of an existing container.
+The easiest way to start writing our own Docker images is by layering on top of existing containers.
 - Remember: [Like ogres](https://www.youtube.com/watch?v=aJQmVZSAqlc), Docker containers are all about layers.
 
 --
 
-Let's see a simple example where we add an R library to our `tidyverse:4.0.0` image. First, make sure that the container is running again:
+Let's see a simple example where we add an R library to our `tidyverse:4.0.0` image. First, make sure that the container is still running. You should see something like:
 
 ```{bash}
-docker ps ## Check that the tidyverse:4.0.0 container isn't still running
-docker run -d -p 8787:8787 -e PASSWORD=pswd123 rocker/tidyverse:4.0.0 ## Run it
+docker ps
+```
+```bash
+CONTAINER ID   IMAGE                    COMMAND   CREATED         STATUS         PORTS                                       NAMES
+802dbd3841c7   rocker/tidyverse:4.0.0   "/init"   8 minutes ago   Up 8 minutes   0.0.0.0:8787->8787/tcp, :::8787->8787/tcp   sweet_maxwell
+
 ```
 
+*(If you don't see something like the above, please re-start your container and then log-in to RStudio Server, using the same steps that we saw [previously](#tverseinit).)*
+
 --
 
-Then log into RStudio Server through your browser, exactly like we did [earlier](#login). Once you are in RStudio, install **data.table** like you would any normal R library.
+Once you are in RStudio, install **data.table** like you would any normal R library.
 - I'm not going to show this with a GIF, but either use RStudio's library installer or run `install.packages("data.table")`.
 
 ---
 
 # Add to an existing container (cont.)
 
-You should now have data.table installed on your running container.
+Okay, data.table should now be installed on your running container.
 
 **Question:** If you stopped your container and restarted it, would data.table still be there?
 
@@ -575,14 +583,14 @@ I'm going to show you how on the next slide to keep everything in one place...
 
 # Add to an existing container (cont.)
 
-**Step 1:** Run `docker ps` again to ID the running container. You should see something like:
+**Step 1:** Run `docker ps` to ID the running container. We've already done this, but...
 
 ```{bash}
 docker ps
 ```
-```bash  
+```bash
 CONTAINER ID   IMAGE                    COMMAND   CREATED         STATUS         PORTS                                       NAMES
-802dbd3841c7   rocker/tidyverse:4.0.0   "/init"   2 minutes ago   Up 2 minutes   0.0.0.0:8787->8787/tcp, :::8787->8787/tcp   sweet_maxwell
+802dbd3841c7   rocker/tidyverse:4.0.0   "/init"   8 minutes ago   Up 8 minutes   0.0.0.0:8787->8787/tcp, :::8787->8787/tcp   sweet_maxwell
 
 ```
 --
@@ -634,32 +642,92 @@ root@802dbd3841c7:/# htop <span class="hljs-comment">## Show all available CPU c
 
 --
 
-(Obviously, you'd have to commit this change to keep it.)
+(Obviously, you'd now have to commit this change to add `htop` to your image.)
 
 ---
 
-# Write your own Dockerfile(s)
+# Aside: Stop your container(s)
+
+Okay, now is a good time to stop your container if you haven't done so already. Grab your container ID and run:
+
+```{bash}
+docker stop <container-id>
+```
+
+Alternatively, you can stop all running containers with the following command:
+
+```{bash}
+docker stop $(docker ps -q)
+```
+
+---
 
-Recall that Dockerfiles are the "sheet music" of the whole operation. These are simple text files that provide the full set of instructions for building our Docker images.
+# Write your own Dockerfile
+
+Recall that `Dockerfiles` are the "sheet music" of the whole operation. These are simple text files that provide the full set of (shell) instructions for building our Docker images.
 
 --
 
-There is a whole host of [commands and considerations](https://docs.docker.com/develop/develop-images/dockerfile_best-practices/) for writing your own Dockerfiles... all of which I am going to elide over for this lecture. (We simply don't have the time.)
+There is a whole host of [commands and considerations](https://docs.docker.com/develop/develop-images/dockerfile_best-practices/) for writing your own Dockerfiles &mdash; all of which I am going to skip for this lecture. (We simply don't have the time.)
 
 --
 
-BUT... I will briefly say that the [Rocker Project](https://github.com/rocker-org/rocker-versioned2#modifying-and-extending-images-in-the-new-architecture) again has our back with a bunch of ready-made scripts for building on and extending their Docker images. 
+BUT... I will briefly say that the Rocker Project again has our backs with a bunch of [ready-made scripts](https://github.com/rocker-org/rocker-versioned2#modifying-and-extending-images-in-the-new-architecture) for building on and extending their Docker images. 
 
 --
 
-For example, if we wanted to modify the `tidyverse:4.0.0` image so that it also included Python, our Dockerfile would be as simple as the following two lines:
+For example, if we wanted to modify the `r-ver4.0.0` image so that it also included Python, our Dockerfile would be as simple as the following two lines:
 
 ```docker
-FROM rocker/tidyverse:4.0.0
+FROM rocker/r-ver:4.0.0
+RUN /rocker_scripts/install_python.sh
+```
+</br></br>
+
+.pull-right[*Continues on next slide.*]
 
+---
+
+# Write your own Dockerfile (cont.)
+
+```docker
+FROM rocker/r-ver:4.0.0
 RUN /rocker_scripts/install_python.sh
 ```
+--
+
+Try this yourself by creating a file called `Dockerfile`<sup>†</sup> comprising the above lines.
+
+.footnote[<sup>†</sup>Every `Dockerfile` is called exactly that. Only one `Dockerfile` is allowed per (sub) directory.]
+
+--
+
+Next, build your Docker image from this `Dockerfile` using the following shell command. I'm going to call my image `r_py` and give it the "4.0.0" version stamp (both choices being optional). **Important:** Make sure that your shell/terminal is in the same directory as the `Dockerfile`when you run this command.
+
+```{bash}
+# docker build --tag <name>:<version> <directory>
+docker build --tag r_py:4.0.0 . 
+```
+
+--
+
+This will take a minute to pull everything in. But your `r_py` image with be ready for immediate deployment thereafter, and now includes Python and [reticulate](https://rstudio.github.io/reticulate/).
+
+```{bash}
+docker run -it --rm r_py:4.0.0
+```
 
+---
+
+# Docker Hub: Share your Docker images
+
+You can share your Dockerfiles and images in various ways.
+
+- I sometimes provide Dockerfiles on the GitHub repo associated with a particular research project. This provides a convenient way for others to reproduce the same computing environment that I used for conducting my analysis. (Example [here](https://github.com/grantmcdermott/sceptic-priors#docker).)
+
+
+The most popular way to share Docker images is by hosting them on [**Docker Hub**](https://hub.docker.com/).
+- I'm not going to show you how to do that here. But the good news is that it's very straightforward. See [here](https://jsta.github.io/r-docker-tutorial/04-Dockerhub.html) for a quick walkthrough.
 
 ---
 class: inverse, center, middle
@@ -670,6 +738,21 @@ name: share
 
 ---
 
+# Prep: Stop all running containers
+
+
+*This next section is all about sharing files and folders between your computer and a container. To avoid unexpected behaviour, it would be best to stop all running containers before continuing.*
+
+```{bash}
+docker stop $(docker ps -q)
+```
+
+--
+
+*You're good to continue now...*
+
+---
+
 # Share files by mounting volumes
 
 Each container runs in a sandboxed environment and cannot access other files and directories on your computer unless you give it explicit permission.
@@ -861,7 +944,7 @@ R users are spoilt, thanks to the Rocker Project. Easy to build our own Dockerfi
 docker run -it --rm rocker/r-ver:4.0.0
 ```
 
-(See next page for a list of key commands.)
+*(See next slide for a list of key commands.)*
 
 ---
 
@@ -877,7 +960,7 @@ docker run -it --rm rocker/r-ver:4.0.0
 
 - `docker ps` list of currently running containers
 
-- `docker stop <containerids>` stop one or more running containers
+- `docker stop <container-ids>` stop one or more running containers
 
 - `docker images` list all installed images
 
@@ -900,14 +983,7 @@ docker run -it --rm rocker/r-ver:4.0.0
 - [Using Docker for Data Science](https://www.robertmylesmcdonnell.com/content/posts/docker/) (Very thorough walkthrough, with a focus on composing your own Dockerfiles from scratch.)
 - [ROpenSci Docker Tutorial](http://ropenscilabs.github.io/r-docker-tutorial) (Another detailed and popular tutorial, albeit outdated in parts.)
 
----
-class: inverse, center, middle
-
-# Next class: Google Compute Engine
-<html><div style='float:left'></div><hr color='#EB811B' size=1px width=796px></html>
-
-
-```{r gen_pdf, include = FALSE, cache = FALSE, eval = FALSE}
+```{r gen_pdf, include = FALSE, cache = FALSE, eval = TRUE}
 infile = list.files(pattern = '.html')
 pagedown::chrome_print(input = infile, timeout = 100)
 ```