Docker Updates #308

jssmith · 2017-02-22T06:37:05Z

This change helps improve Docker usability:

Fixes broken paths in Dockerfile
Use git archive to export Ray sources to container
start_ray.sh does not exit when used within Docker
Updated "Install on Docker" instructions
Created "Using Ray on a Docker Cluster" instructions

Note that this change does not include presently linking to Docker instructions from the top-level README. This is something we should do as soon as we have confidence the instructions work smoothly for people.

robertnishihara · 2017-02-22T06:39:17Z

We should link to these instructions from README.md, right?

AmplabJenkins · 2017-02-22T06:42:27Z

Merged build finished. Test PASSed.

AmplabJenkins · 2017-02-22T06:42:28Z

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/69/
Test PASSed.

jssmith · 2017-02-22T06:42:34Z

@robertnishihara, ah, I was just updating the description. I'd like to see a few people try it before we link to the README.md, but perhaps we can pass that mark in this code review cycle.

robertnishihara

Looks great! I haven't had a chance to try out the cluster version yet.

robertnishihara · 2017-02-23T03:22:54Z

doc/install-on-docker.md

-You can install Ray on any platform that runs Docker. We do not presently publish Docker images for Ray, but you can build them yourself using the Ray distribution. Using Docker can provide a reliable way to get up and running quickly.
+You can install Ray on any platform that runs Docker. We do not presently publish Docker images for Ray, but you can build them yourself using the Ray distribution.
+
+Using Docker can streamline the build process reliable way to get up and running quickly.


"the build process in a reliable way"?

robertnishihara · 2017-02-23T03:24:07Z

doc/install-on-docker.md

-The Docker Platform release is available for Mac, Windows, and Linux platforms. Please download the appropriate version from the [Docker website](https://www.docker.com/products/overview#/install_the_platform).
+### Mac, Linux, Windows platforms
+
+The Docker Platform release is available for Mac, Windows, and Linux platforms. Please download the appropriate version from the [Docker website](https://www.docker.com/products/overview#/install_the_platform) and follow the corresponding installation instructions.


I remember thinking/hearing that the official docker installation instructions for Ubuntu are bad and that the ones to follow are here https://www.digitalocean.com/community/tutorials/how-to-install-and-use-docker-on-ubuntu-16-04

Have you ever had any trouble with the official docker instructions?

Yeah, the official Linux instructions are not very relevant to our use case. I'll add a link to the alternate instructions.

robertnishihara · 2017-02-23T03:30:33Z

doc/install-on-docker.md

+
+The Docker Platform release is available for Mac, Windows, and Linux platforms. Please download the appropriate version from the [Docker website](https://www.docker.com/products/overview#/install_the_platform) and follow the corresponding installation instructions.
+
+### Docker installation on EC2 with Ubuntu


These instructions are different from the ones on the Docker website. E.g., the official Docker instructions do not do sudo apt-get -y dist-upgrade.

Why the difference?

The official Linux instructions seem to focus on connecting to Dockerhub and getting a repository set up. These are simple instructions to get Docker installed and to get it to run without sudo.

The dist-upgrade command here is likely optional. It' something I always do just so I'm running the latest patched version of the OS. I know things have at times broken without it but I don't remember whether it is necessary right now. I think it's probably safest to leave it in here.

robertnishihara · 2017-02-23T03:33:58Z

doc/install-on-docker.md


 ## Launch Ray in Docker

 Start out by launching the deployment container.

 ```
-docker run --shm-size=1024m -t -i ray-project/ray:deploy
+docker run --shm-size=<shm-size> -t -i ray-project/ray:deploy


This command doesn't work, I think you need to remove the ray: and just use ray-project/deploy

robertnishihara · 2017-02-23T03:36:45Z

doc/install-on-docker.md

 ## Test if the installation succeeded

-To test if the installation was successful, try running some tests.
+To test if the installation was successful, try running some tests. Within the container shell enter the following commands:

 ```
 python test/runtest.py # This tests basic functionality.


I just tried this and it told me

Traceback (most recent call last): File "test/runtest.py", line 127, in <module> raise Exception("You have an older version of cloudpickle that is not able to serialize namedtuples. Try running \n\n{}\n\n".format(cloudpickle_command)) Exception: You have an older version of cloudpickle that is not able to serialize namedtuples. Try running pip install --upgrade cloudpickle

We should make sure that we're installing the correct version of cloudpickle in the docker image. For that it should be sufficient to do pip install cloudpickle somewhere.

robertnishihara · 2017-02-23T03:46:31Z

doc/using-ray-and-docker-on-a-cluster.md

+FROM ray-project/deploy
+RUN git clone https://github.com/my-user/my-project.git
+RUN ./my-project/install.sh
+```


I don't understand this example. What is my-project supposed to be? And where would the installation of Ray happen?

robertnishihara · 2017-02-23T03:48:09Z

doc/using-ray-and-docker-on-a-cluster.md

+    <repository-uri> \
+    /ray/scripts/start_ray.sh --head \
+        --redis-port=6379 \
+        --num-workers=<num-workers>


shall we also have people use a fixed object store port?

Both here and on the worker nodes

Is there a port number we have standardized on?

Not yet. Maybe just choose one? E.g., 7183.

robertnishihara · 2017-02-23T03:51:49Z

doc/install-on-docker.md

@@ -59,7 +126,7 @@ docker run --shm-size=1024m -t -i ray-project/ray:examples



the command above

docker run --shm-size=1024m -t -i ray-project/ray:examples

should be

docker run --shm-size=1024m -t -i ray-project/examples

robertnishihara · 2017-02-23T03:53:39Z

docker/base-deps/Dockerfile

@@ -10,5 +10,5 @@ RUN echo 'export PATH=/opt/conda/bin:$PATH' > /etc/profile.d/conda.sh \
    && /bin/bash /tmp/anaconda.sh -b -p /opt/conda \
    && rm /tmp/anaconda.sh
 ENV PATH "/opt/conda/bin:$PATH"
-RUN conda install libgcc
+RUN conda install -y libgcc
 RUN pip install --upgrade pip


probably want do pip install some stuff (at least cloudpickle) here

Will add cloudpickle.

robertnishihara · 2017-02-23T03:55:42Z

doc/install-on-docker.md

@@ -68,7 +135,7 @@ See the [Hyperparameter optimization documentation](../examples/hyperopt/README.
 ### Batch L-BFGS

 ```
-cd ~/ray/examples/lbfgs/
+cd /ray/examples/lbfgs/


When I run this example I get a lot of errors like the following.

W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available on your machine and could speed up CPU computations. W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations. W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations. W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations. W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations. W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations. [INFO] (/ray/src/photon/photon_scheduler.c:202) Started worker with pid 4463 W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE3 instructions, but these are available on your machine and could speed up CPU computations.

I looked at forums a bit and these warnings seem to come from some recent changes in Tensorflow. I don't believe this is related to Ray or Docker and it something we may want to come back to.

AmplabJenkins · 2017-02-27T23:17:30Z

Merged build finished. Test PASSed.

AmplabJenkins · 2017-02-27T23:17:30Z

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/123/
Test PASSed.

AmplabJenkins · 2017-02-27T23:27:36Z

Merged build finished. Test PASSed.

AmplabJenkins · 2017-02-27T23:27:36Z

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/124/
Test PASSed.

AmplabJenkins · 2017-02-28T01:23:28Z

Merged build finished. Test FAILed.

AmplabJenkins · 2017-02-28T01:23:29Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/127/
Test FAILed.

AmplabJenkins · 2017-02-28T01:43:27Z

Merged build finished. Test FAILed.

AmplabJenkins · 2017-02-28T01:43:27Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/128/
Test FAILed.

AmplabJenkins · 2017-02-28T05:51:09Z

Merged build finished. Test FAILed.

AmplabJenkins · 2017-02-28T05:51:09Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/130/
Test FAILed.

AmplabJenkins · 2017-02-28T07:48:26Z

Merged build finished. Test FAILed.

AmplabJenkins · 2017-02-28T07:48:26Z

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/Ray-PRB/132/
Test FAILed.

robertnishihara · 2017-02-28T19:34:06Z

Btw, I ran the docker instructions without the sudo apt-get -y dist-upgrade and it worked fine.

robertnishihara reviewed Feb 23, 2017

View reviewed changes

jssmith force-pushed the fixdocker2 branch from 0f376a4 to de67028 Compare February 27, 2017 23:11

jssmith added 12 commits February 27, 2017 21:39

new path for python build

93d7873

add flag

db2a526

build tar using git archive

09b9a13

no exit from start_ray.sh

732ee8a

update Docker instructions

27d68d8

add git revision

349244c

update build docker script

6a50682

fix typo

f0e741b

bug fixes and clarifications

32794b4

mend

4a9855c

add objectmanager ports to docker instructions

7a44fe7

rewording

d019ea0

robertnishihara force-pushed the fixdocker2 branch from 81041da to d019ea0 Compare February 28, 2017 05:40

Small updates to documentation.

955f375

robertnishihara merged commit ad4b03b into ray-project:master Mar 1, 2017

robertnishihara deleted the fixdocker2 branch March 1, 2017 02:57


		The Docker Platform release is available for Mac, Windows, and Linux platforms. Please download the appropriate version from the [Docker website](https://www.docker.com/products/overview#/install_the_platform) and follow the corresponding installation instructions.

		### Docker installation on EC2 with Ubuntu

		@@ -59,7 +126,7 @@ docker run --shm-size=1024m -t -i ray-project/ray:examples

Docker Updates #308

Docker Updates #308

Conversation

jssmith commented Feb 22, 2017 • edited Loading

robertnishihara commented Feb 22, 2017

AmplabJenkins commented Feb 22, 2017

AmplabJenkins commented Feb 22, 2017

jssmith commented Feb 22, 2017

robertnishihara left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

AmplabJenkins commented Feb 27, 2017

AmplabJenkins commented Feb 27, 2017

AmplabJenkins commented Feb 27, 2017

AmplabJenkins commented Feb 27, 2017

AmplabJenkins commented Feb 28, 2017

AmplabJenkins commented Feb 28, 2017

AmplabJenkins commented Feb 28, 2017

AmplabJenkins commented Feb 28, 2017

AmplabJenkins commented Feb 28, 2017

AmplabJenkins commented Feb 28, 2017

AmplabJenkins commented Feb 28, 2017

AmplabJenkins commented Feb 28, 2017

robertnishihara commented Feb 28, 2017

jssmith commented Feb 22, 2017 •

edited

Loading