Cleanup onbuild docker file.

Fix issue with chmod not being able to run in onbuild
- now using python to launch the main file

Smart use of copy (requirements.txt* Dockerfile) to make sure we specific files in docker build process before copying all files. By using requirements.txt* Dockerfile we will make sure we always copy a file (Dockerfile) and the copy operation will not fail.

Added more documentation on how to use onbuild.

Author: robkooper

Dockerfile

@@ -1,5 +1,4 @@
 FROM ubuntu:16.04
-MAINTAINER Rob Kooper <[email protected]>
 
 # environment variables
 ENV RABBITMQ_URI="" \

Dockerfile.onbuild

@@ -1,22 +1,22 @@
 FROM clowder/pyclowder:2
 
-# copy all files
-ONBUILD ADD . /home/clowder/
-
 # install any packages
-#ONBUILD COPY packages.apt /home/clowder/
+ONBUILD COPY packages.* Dockerfile /home/clowder/
 ONBUILD RUN if [ -e packages.apt ]; then \
                 apt-get -q -q update \
                 && xargs apt-get -y install --no-install-recommends < packages.apt \
                 && rm -rf /var/lib/apt/lists/*; \
             fi
 
 # install any python packages
-#ONBUILD COPY requirements.txt /home/clowder/
+ONBUILD COPY requirements.txt* Dockerfile /home/clowder/
 ONBUILD RUN if [ -e requirements.txt ]; then \
                 pip install --no-cache-dir -r requirements.txt; \
             fi
 
+# copy all files
+ONBUILD ADD . /home/clowder/
+
 # switch to user clowder last minute
 ONBUILD USER clowder
 

README.md

@@ -202,4 +202,24 @@ an extractor or code that interacts with clowder. One of these functions is setu
 logging system for you. The logging function takes a single argument that can be None. The argument is either a pointer
 to a file that is read with the configuration options.
 
-# files
+## files
+
+# Dockerfile
+
+We recommend using the pyclowder:onbuild to easily convert your extractor into a docker container. If you build the
+extractor as commented above, you will only need the following Dockerfile
+
+```
+FROM clowder/pyclowder:onbuild
+
+ENV MAIN_SCRIPT="wordcount.py"
+```
+
+The main piece is the MAIN_SCRIPT which should point to the python file that holds your main function.
+
+If you need additional packages installed, you will need a file called packages.apt with in this file a list of all
+packages that need to be installed. The docker build process will use this file to install those pacakges first in
+the docker container.
+
+If you need any python packages installed you will need to create file called requiremenets.txt. If this file exists
+the docker build process will use `pip install -r requirements.txt` to install these packages.

entrypoint.sh

@@ -21,7 +21,6 @@ if [ "$1" = 'extractor' ]; then
         echo "Main script specified does not exist."
         exit -1
     fi
-    chmod 755 "${MAIN_SCRIPT}"
 
     # check to make sure rabbitmq is up
     if [ "${RABBITMQ_PORT_5672_TCP_ADDR}" != "" ]; then
@@ -35,7 +34,7 @@ if [ "$1" = 'extractor' ]; then
     fi
 
     # launch extractor and see what happens
-    exec "./${MAIN_SCRIPT}"
+    exec "python ./${MAIN_SCRIPT}"
 fi
 
 exec "$@"

sample-extractors/wordcount/Dockerfile

@@ -1,4 +1,3 @@
 FROM clowder/pyclowder:onbuild
 
-ENV RABBITMQ_QUEUE="ncsa.wordcount" \
-    MAIN_SCRIPT="wordcount.py"
+ENV MAIN_SCRIPT="wordcount.py"