You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by yanboliang <gi...@git.apache.org> on 2016/12/08 13:14:29 UTC
[GitHub] spark pull request #16214: [[SPARK-18325]][SPARKR] Add example for using nat...
GitHub user yanboliang opened a pull request:
https://github.com/apache/spark/pull/16214
[[SPARK-18325]][SPARKR] Add example for using native R package.
## What changes were proposed in this pull request?
SparkR provides the ability to distribute the computations by ```spark.lappy```. In many cases, the distributed computation depends on third-party R packages and we need to install corresponding packages to executor in advance. Since SparkR is an interactive analytic tools and users may use lots of native R packages across the session, so I think it's helpful to provide an example to illustrate how to handle this scenario.
## How was this patch tested?
Offline test.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/yanboliang/spark native-r-package
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/spark/pull/16214.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #16214
----
commit 9b15bb3ccf1b293696f166bb1d64b96906ecbbf9
Author: Yanbo Liang <yb...@gmail.com>
Date: 2016-12-07T15:04:03Z
Add example for using native R package.
----
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91881006
--- Diff: docs/sparkr.md ---
@@ -472,21 +472,17 @@ should fit in a single machine. If that is not the case they can do something li
`dapply`
<div data-lang="r" markdown="1">
-{% highlight r %}
-# Perform distributed training of multiple models with spark.lapply. Here, we pass
-# a read-only list of arguments which specifies family the generalized linear model should be.
--- End diff --
Sounds good, updated.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang closed the pull request at:
https://github.com/apache/spark/pull/16214
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by AmplabJenkins <gi...@git.apache.org>.
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/16214
Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/69908/
Test PASSed.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by AmplabJenkins <gi...@git.apache.org>.
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/16214
Merged build finished. Test PASSed.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by AmplabJenkins <gi...@git.apache.org>.
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/16214
Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/70000/
Test PASSed.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by AmplabJenkins <gi...@git.apache.org>.
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/16214
Merged build finished. Test PASSed.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by SparkQA <gi...@git.apache.org>.
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/16214
**[Test build #69908 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/69908/consoleFull)** for PR 16214 at commit [`ce18e2e`](https://github.com/apache/spark/commit/ce18e2eb181ddf6079640049e2ed2d18bb8ae03c).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91667478
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,80 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to use third-party R packages in your task
+# which is distributed by Spark. We support two scenarios:
+# - Install packages from CRAN to executors directly.
+# - Install packages from local file system to executors.
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# Get the location of the default library
+libDir <- .libPaths()[1]
--- End diff --
Good suggestion!
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by AmplabJenkins <gi...@git.apache.org>.
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/16214
Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/69861/
Test PASSed.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by SparkQA <gi...@git.apache.org>.
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/16214
**[Test build #70000 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/70000/consoleFull)** for PR 16214 at commit [`d3ec5fa`](https://github.com/apache/spark/commit/d3ec5fabf686c4a96a5032d716e5ef1eff7fb8c1).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on the issue:
https://github.com/apache/spark/pull/16214
Great - I think we should be careful with example code that install stuff automatically, especially if that could run on a cluster without any change.
Perhaps we should have a blocking prompt at the beginning with a warning message?
Also it would be better to have more description on what it is doing, where things would go (eg. where package would be installed to), and why?
And perhaps a section on programming guide? Or that is too public (depending on how much we want to promote such things)?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by AmplabJenkins <gi...@git.apache.org>.
Github user AmplabJenkins commented on the issue:
https://github.com/apache/spark/pull/16214
Merged build finished. Test PASSed.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91562472
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,80 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to use third-party R packages in your task
+# which is distributed by Spark. We support two scenarios:
+# - Install packages from CRAN to executors directly.
+# - Install packages from local file system to executors.
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# Get the location of the default library
+libDir <- .libPaths()[1]
+
+# Install third-party R packages from CRAN to executors directly if it does not exist,
+# then the packages can be used by the corresponding task.
+
+# Perform distributed training of multiple models with spark.lapply
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(libDir)) == FALSE) {
+ install.packages("e1071", repos = "https://cran.r-project.org")
+ }
+ library(e1071)
+ model <- svm(Species ~ ., data = iris, cost = cost)
+ summary(model)
+}
+model.summaries <- spark.lapply(costs, train)
+
+# Print the summary of each model
+print(model.summaries)
+
+# Install third-party R packages from local file system to executors if it does not exist,
+# then the packages can be used by the corresponding task.
+
+# Downloaded e1071 package source code to a directory
+packagesDir <- paste0(tempdir(), "/", "packages")
+dir.create(packagesDir)
+download.packages("e1071", packagesDir, repos = "https://cran.r-project.org")
+filename <- list.files(packagesDir, "^e1071")
+packagesPath <- file.path(packagesDir, filename)
+# Add the third-party R package to be downloaded with this Spark job on every node.
+spark.addFile(packagesPath)
+
+path <- spark.getSparkFiles(filename)
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(libDir)) == FALSE) {
+ install.packages(path, repos=NULL, type="source")
--- End diff --
https://stat.ethz.ch/R-manual/R-devel/library/base/html/library.html
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91766006
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,68 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to install third-party R packages to executors
+# in your SparkR jobs distributed by "spark.lapply".
+#
+# Note: This example will install packages to a temporary directory on your machine.
+# The directory will be removed automatically when the example exit.
+# You environment should be connected to internet to run this example,
+# otherwise, you should change "repos" to your private repository url.
+# And the environment need to have necessary tools such as gcc to compile
+# and install R package "e1071".
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# $example on$
+# The directory where the third-party R packages are installed.
+libDir <- paste0(tempdir(), "/", "Rlib")
+dir.create(libDir)
+
+# Downloaded e1071 package source code to a directory
+packagesDir <- paste0(tempdir(), "/", "packages")
+dir.create(packagesDir)
+download.packages("e1071", packagesDir, repos = "https://cran.r-project.org")
+filename <- list.files(packagesDir, "^e1071")
+packagesPath <- file.path(packagesDir, filename)
+# Add the third-party R package to be downloaded with this Spark job on every node.
+spark.addFile(packagesPath)
+
+path <- spark.getSparkFiles(filename)
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(lib = libDir)) == FALSE) {
+ install.packages(path, repos = NULL, type = "source")
--- End diff --
re: comment [here](https://github.com/apache/spark/pull/16214#discussion_r91562425) if you already have the package content from sparkFiles you do not need to call `install.packages()`, which I think would be better without it.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by SparkQA <gi...@git.apache.org>.
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/16214
**[Test build #69861 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/69861/consoleFull)** for PR 16214 at commit [`9b15bb3`](https://github.com/apache/spark/commit/9b15bb3ccf1b293696f166bb1d64b96906ecbbf9).
* This patch passes all tests.
* This patch merges cleanly.
* This patch adds no public classes.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by SparkQA <gi...@git.apache.org>.
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/16214
**[Test build #69861 has started](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/69861/consoleFull)** for PR 16214 at commit [`9b15bb3`](https://github.com/apache/spark/commit/9b15bb3ccf1b293696f166bb1d64b96906ecbbf9).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91765571
--- Diff: docs/sparkr.md ---
@@ -472,21 +472,17 @@ should fit in a single machine. If that is not the case they can do something li
`dapply`
<div data-lang="r" markdown="1">
-{% highlight r %}
-# Perform distributed training of multiple models with spark.lapply. Here, we pass
-# a read-only list of arguments which specifies family the generalized linear model should be.
-families <- c("gaussian", "poisson")
-train <- function(family) {
- model <- glm(Sepal.Length ~ Sepal.Width + Species, iris, family = family)
- summary(model)
-}
-# Return a list of model's summaries
-model.summaries <- spark.lapply(families, train)
+{% include_example lapply r/ml/ml.R %}
+</div>
-# Print the summary of each model
-print(model.summaries)
+##### spark.lapply with third-party packages
-{% endhighlight %}
+Many of the SparkR jobs distributed by `spark.lapply` need supports from third-party packages. Rather than installing all necessary packages to all executors in advance,
+we could install them during the SparkR interactive session or script. Users can add package files or directories by `spark.addFile` firstly,
+download them to every executor node, and install them.
--- End diff --
this kind of sounds like the user will need to separately "download them to executor node" - perhaps instead say "by `spark.addFile` first, which automatically download them to every executor node, and then install them`?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91679010
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,68 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to install third-party R packages to executors
+# in your SparkR jobs distributed by "spark.lapply".
+#
+# Note: This example will install packages to a temporary directory on your machine.
+# The directory will be removed automatically when the example exit.
+# You environment should be connected to internet to run this example,
+# otherwise, you should change "repos" to your private repository url.
+# And the environment need to have necessary tools such as gcc to compile
+# and install R package "e1071".
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# $example on$
+# The directory where the third-party R packages are installed.
+libDir <- paste0(tempdir(), "/", "Rlib")
+dir.create(libDir)
+
+# Downloaded e1071 package source code to a directory
+packagesDir <- paste0(tempdir(), "/", "packages")
+dir.create(packagesDir)
+download.packages("e1071", packagesDir, repos = "https://cran.r-project.org")
+filename <- list.files(packagesDir, "^e1071")
+packagesPath <- file.path(packagesDir, filename)
+# Add the third-party R package to be downloaded with this Spark job on every node.
+spark.addFile(packagesPath)
+
+path <- spark.getSparkFiles(filename)
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(lib = libDir)) == FALSE) {
+ install.packages(path, repos = NULL, type = "source")
+ }
+ library(e1071)
--- End diff --
Yeah, I run it in standalone mode several times, all work well. But I think I can not guarantee it works well always before more careful test, may be I'm lucky and not hit the concurrent issue. I'll figure out a religious test to verify it later. Thanks.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang commented on the issue:
https://github.com/apache/spark/pull/16214
@mengxr I think there is no (at least I did not found) concurrency issue when multiple executors on the same machine, since native R has lock to protect this. You can verify it by installing package to a shared directory in a single node in multi-thread way. However, we should recommend to install package to an executor associated directory which means different executor has its own R lib directory even if they are in the same machine. For YARN mode, it works well since each executor has its own R lib directory natively.
@felixcheung I updated the examples and add corresponding session in SparkR user guide. I'm not sure whether such an example is appropriate, since it has many dependencies on environment. But I think for an interactive analytical tool, installing packages across the session is not very rare. Still, I'm open to hear your thought, thanks.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang commented on the issue:
https://github.com/apache/spark/pull/16214
After rethinking this issue, I found this is not very appropriate as an example delivered with Spark, since it depends on the environment that users running the script. I'm closing this PR, thanks for all your reviewing.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91765125
--- Diff: docs/sparkr.md ---
@@ -472,21 +472,17 @@ should fit in a single machine. If that is not the case they can do something li
`dapply`
<div data-lang="r" markdown="1">
-{% highlight r %}
-# Perform distributed training of multiple models with spark.lapply. Here, we pass
-# a read-only list of arguments which specifies family the generalized linear model should be.
-families <- c("gaussian", "poisson")
-train <- function(family) {
- model <- glm(Sepal.Length ~ Sepal.Width + Species, iris, family = family)
- summary(model)
-}
-# Return a list of model's summaries
--- End diff --
ditto, but minor here
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang commented on the issue:
https://github.com/apache/spark/pull/16214
@felixcheung @shivaram
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by mengxr <gi...@git.apache.org>.
Github user mengxr commented on the issue:
https://github.com/apache/spark/pull/16214
@yanboliang What happens if there are multiple executors on the same machine? Are there concurrency issues?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by SparkQA <gi...@git.apache.org>.
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/16214
**[Test build #69908 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/69908/consoleFull)** for PR 16214 at commit [`ce18e2e`](https://github.com/apache/spark/commit/ce18e2eb181ddf6079640049e2ed2d18bb8ae03c).
* This patch passes all tests.
* This patch merges cleanly.
* This patch adds no public classes.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark issue #16214: [SPARK-18325][SPARKR] Add example for using native R pac...
Posted by SparkQA <gi...@git.apache.org>.
Github user SparkQA commented on the issue:
https://github.com/apache/spark/pull/16214
**[Test build #70000 has finished](https://amplab.cs.berkeley.edu/jenkins/job/SparkPullRequestBuilder/70000/consoleFull)** for PR 16214 at commit [`d3ec5fa`](https://github.com/apache/spark/commit/d3ec5fabf686c4a96a5032d716e5ef1eff7fb8c1).
* This patch passes all tests.
* This patch merges cleanly.
* This patch adds no public classes.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91562425
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,80 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to use third-party R packages in your task
+# which is distributed by Spark. We support two scenarios:
+# - Install packages from CRAN to executors directly.
+# - Install packages from local file system to executors.
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# Get the location of the default library
+libDir <- .libPaths()[1]
+
+# Install third-party R packages from CRAN to executors directly if it does not exist,
+# then the packages can be used by the corresponding task.
+
+# Perform distributed training of multiple models with spark.lapply
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(libDir)) == FALSE) {
+ install.packages("e1071", repos = "https://cran.r-project.org")
+ }
+ library(e1071)
+ model <- svm(Species ~ ., data = iris, cost = cost)
+ summary(model)
+}
+model.summaries <- spark.lapply(costs, train)
+
+# Print the summary of each model
+print(model.summaries)
+
+# Install third-party R packages from local file system to executors if it does not exist,
+# then the packages can be used by the corresponding task.
+
+# Downloaded e1071 package source code to a directory
+packagesDir <- paste0(tempdir(), "/", "packages")
+dir.create(packagesDir)
+download.packages("e1071", packagesDir, repos = "https://cran.r-project.org")
+filename <- list.files(packagesDir, "^e1071")
+packagesPath <- file.path(packagesDir, filename)
+# Add the third-party R package to be downloaded with this Spark job on every node.
+spark.addFile(packagesPath)
+
+path <- spark.getSparkFiles(filename)
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(libDir)) == FALSE) {
+ install.packages(path, repos=NULL, type="source")
--- End diff --
although if this is an example of how R package can be distributed, I wouldn't call `install.packages` here, because of secure location and duplications.
instead, this could do `library(e1071, lib.loc = path)` - ie. the package doesn't need to be "installed" to be loaded.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91561800
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,80 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to use third-party R packages in your task
+# which is distributed by Spark. We support two scenarios:
+# - Install packages from CRAN to executors directly.
+# - Install packages from local file system to executors.
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# Get the location of the default library
+libDir <- .libPaths()[1]
+
+# Install third-party R packages from CRAN to executors directly if it does not exist,
+# then the packages can be used by the corresponding task.
+
+# Perform distributed training of multiple models with spark.lapply
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(libDir)) == FALSE) {
--- End diff --
I'd prefer `installed.packages(lib = libDir)` to be more clear
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by yanboliang <gi...@git.apache.org>.
Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91881854
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,68 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to install third-party R packages to executors
+# in your SparkR jobs distributed by "spark.lapply".
+#
+# Note: This example will install packages to a temporary directory on your machine.
+# The directory will be removed automatically when the example exit.
+# You environment should be connected to internet to run this example,
+# otherwise, you should change "repos" to your private repository url.
+# And the environment need to have necessary tools such as gcc to compile
+# and install R package "e1071".
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# $example on$
+# The directory where the third-party R packages are installed.
+libDir <- paste0(tempdir(), "/", "Rlib")
+dir.create(libDir)
+
+# Downloaded e1071 package source code to a directory
+packagesDir <- paste0(tempdir(), "/", "packages")
+dir.create(packagesDir)
+download.packages("e1071", packagesDir, repos = "https://cran.r-project.org")
+filename <- list.files(packagesDir, "^e1071")
+packagesPath <- file.path(packagesDir, filename)
+# Add the third-party R package to be downloaded with this Spark job on every node.
+spark.addFile(packagesPath)
+
+path <- spark.getSparkFiles(filename)
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(lib = libDir)) == FALSE) {
+ install.packages(path, repos = NULL, type = "source")
--- End diff --
Yeah, we have the package content, but it's source package rather than binary package, so we can not use ```library``` to load the package. This is the pain point for this example. If we illustrate this example with binary package, we should provide scripts for different os version, and it require all nodes in users' cluster should have the same architecture. So I use source package, I think it's a more universal example.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
[GitHub] spark pull request #16214: [SPARK-18325][SPARKR] Add example for using nativ...
Posted by felixcheung <gi...@git.apache.org>.
Github user felixcheung commented on a diff in the pull request:
https://github.com/apache/spark/pull/16214#discussion_r91561992
--- Diff: examples/src/main/r/native-r-package.R ---
@@ -0,0 +1,80 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements. See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# This example illustrates how to use third-party R packages in your task
+# which is distributed by Spark. We support two scenarios:
+# - Install packages from CRAN to executors directly.
+# - Install packages from local file system to executors.
+#
+# To run this example use
+# ./bin/spark-submit examples/src/main/r/native-r-package.R
+
+# Load SparkR library into your R session
+library(SparkR)
+
+# Initialize SparkSession
+sparkR.session(appName = "SparkR-native-r-package-example")
+
+# Get the location of the default library
+libDir <- .libPaths()[1]
+
+# Install third-party R packages from CRAN to executors directly if it does not exist,
+# then the packages can be used by the corresponding task.
+
+# Perform distributed training of multiple models with spark.lapply
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(libDir)) == FALSE) {
+ install.packages("e1071", repos = "https://cran.r-project.org")
+ }
+ library(e1071)
+ model <- svm(Species ~ ., data = iris, cost = cost)
+ summary(model)
+}
+model.summaries <- spark.lapply(costs, train)
+
+# Print the summary of each model
+print(model.summaries)
+
+# Install third-party R packages from local file system to executors if it does not exist,
+# then the packages can be used by the corresponding task.
+
+# Downloaded e1071 package source code to a directory
+packagesDir <- paste0(tempdir(), "/", "packages")
+dir.create(packagesDir)
+download.packages("e1071", packagesDir, repos = "https://cran.r-project.org")
+filename <- list.files(packagesDir, "^e1071")
+packagesPath <- file.path(packagesDir, filename)
+# Add the third-party R package to be downloaded with this Spark job on every node.
+spark.addFile(packagesPath)
+
+path <- spark.getSparkFiles(filename)
+costs <- exp(seq(from = log(1), to = log(1000), length.out = 5))
+train <- function(cost) {
+ if("e1071" %in% rownames(installed.packages(libDir)) == FALSE) {
--- End diff --
ditto
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org