diff --git a/.circleci/config.yml b/.circleci/config.yml
index 2374e59e6a886..43f2d58acdf31 100644
--- a/.circleci/config.yml
+++ b/.circleci/config.yml
@@ -2,7 +2,7 @@ version: 2
 
 defaults: &defaults
   docker:
-    - image: palantirtechnologies/circle-spark-base
+    - image: palantirtechnologies/circle-spark-base:0.1.0
   resource_class: xlarge
   environment: &defaults-environment
     TERM: dumb
@@ -128,7 +128,7 @@ jobs:
     <<: *defaults
     # Some part of the maven setup fails if there's no R, so we need to use the R image here
     docker:
-      - image: palantirtechnologies/circle-spark-r
+      - image: palantirtechnologies/circle-spark-r:0.1.0
     steps:
       # Saves us from recompiling every time...
       - restore_cache:
@@ -300,7 +300,7 @@ jobs:
     # depends on build-sbt, but we only need the assembly jars
     <<: *defaults
     docker:
-      - image: palantirtechnologies/circle-spark-python
+      - image: palantirtechnologies/circle-spark-python:0.1.0
     parallelism: 2
     steps:
       - *checkout-code
@@ -325,7 +325,7 @@ jobs:
     # depends on build-sbt, but we only need the assembly jars
     <<: *defaults
     docker:
-      - image: palantirtechnologies/circle-spark-r
+      - image: palantirtechnologies/circle-spark-r:0.1.0
     steps:
       - *checkout-code
       - attach_workspace:
@@ -438,7 +438,7 @@ jobs:
     <<: *defaults
     # Some part of the maven setup fails if there's no R, so we need to use the R image here
     docker:
-      - image: palantirtechnologies/circle-spark-r
+      - image: palantirtechnologies/circle-spark-r:0.1.0
     steps:
       - *checkout-code
       - restore_cache:
@@ -458,7 +458,7 @@ jobs:
   deploy-gradle:
     <<: *defaults
     docker:
-      - image: palantirtechnologies/circle-spark-r
+      - image: palantirtechnologies/circle-spark-r:0.1.0
     steps:
       - *checkout-code
       - *restore-gradle-wrapper-cache
@@ -470,7 +470,7 @@ jobs:
     <<: *defaults
     # Some part of the maven setup fails if there's no R, so we need to use the R image here
     docker:
-      - image: palantirtechnologies/circle-spark-r
+      - image: palantirtechnologies/circle-spark-r:0.1.0
     steps:
       # This cache contains the whole project after version was set and mvn package was called
       # Restoring first (and instead of checkout) as mvn versions:set mutates real source code...
diff --git a/dev/docker-images/Makefile b/dev/docker-images/Makefile
index eb1692f810620..ed3e3a5ee7687 100644
--- a/dev/docker-images/Makefile
+++ b/dev/docker-images/Makefile
@@ -17,9 +17,9 @@
 
 .PHONY: all publish base python r
 
-BASE_IMAGE_NAME = palantirtechnologies/circle-spark-base
-PYTHON_IMAGE_NAME = palantirtechnologies/circle-spark-python
-R_IMAGE_NAME = palantirtechnologies/circle-spark-r
+BASE_IMAGE_NAME = palantirtechnologies/circle-spark-base:0.1.0
+PYTHON_IMAGE_NAME = palantirtechnologies/circle-spark-python:0.1.0
+R_IMAGE_NAME = palantirtechnologies/circle-spark-r:0.1.0
 
 all: base python r
 
diff --git a/dev/docker-images/base/Dockerfile b/dev/docker-images/base/Dockerfile
index e4515ca510b23..0e84ec665fcd7 100644
--- a/dev/docker-images/base/Dockerfile
+++ b/dev/docker-images/base/Dockerfile
@@ -15,7 +15,7 @@
 # limitations under the License.
 #
 
-FROM buildpack-deps:xenial
+FROM buildpack-deps:cosmic
 
 # make Apt non-interactive
 RUN echo 'APT::Get::Assume-Yes "true";' > /etc/apt/apt.conf.d/90circleci \
@@ -107,10 +107,11 @@ WORKDIR $CIRCLE_HOME
 # Install miniconda, we are using it to test conda support and a bunch of tests expect CONDA_BIN to be set
 ENV CONDA_ROOT=$CIRCLE_HOME/miniconda
 ENV CONDA_BIN=$CIRCLE_HOME/miniconda/bin/conda
-ENV MINICONDA2_VERSION=4.3.31
+ENV MINICONDA2_VERSION=4.5.11
 RUN curl -sO https://repo.continuum.io/miniconda/Miniconda2-${MINICONDA2_VERSION}-Linux-x86_64.sh \
   && bash Miniconda2-${MINICONDA2_VERSION}-Linux-x86_64.sh -b -p ${CONDA_ROOT} \
   && $CONDA_BIN clean --all \
+  && sudo mkdir -m 777 /home/.conda \
   && rm -f Miniconda2-${MINICONDA2_VERSION}-Linux-x86_64.sh
 
 # END IMAGE CUSTOMIZATIONS
diff --git a/dev/docker-images/python/Dockerfile b/dev/docker-images/python/Dockerfile
index d5674357cce69..cb44b373617da 100644
--- a/dev/docker-images/python/Dockerfile
+++ b/dev/docker-images/python/Dockerfile
@@ -21,15 +21,14 @@ FROM palantirtechnologies/circle-spark-base
 ENV PATH="$CIRCLE_HOME/.pyenv/bin:$PATH"
 RUN curl -L https://github.com/pyenv/pyenv-installer/raw/master/bin/pyenv-installer | bash \
   && cat >>.bashrc <<<'eval "$($HOME/.pyenv/bin/pyenv init -)"' \
-  && cat >>.bashrc <<<'eval "$($HOME/.pyenv/bin/pyenv virtualenv-init -)"' \
-  && pyenv doctor
+  && cat >>.bashrc <<<'eval "$($HOME/.pyenv/bin/pyenv virtualenv-init -)"'
 
 # Must install numpy 1.11 or else a bunch of tests break due to different output formatting on e.g. nparray
 # A version I've tested earlier that I know it breaks with is 1.14.1
 RUN mkdir -p $(pyenv root)/versions \
   && ln -s $CONDA_ROOT $(pyenv root)/versions/our-miniconda \
-  && $CONDA_BIN create -y -n python2 -c anaconda -c conda-forge python==2.7.11 numpy=1.11.2 pyarrow==0.8.0 pandas nomkl \
-  && $CONDA_BIN create -y -n python3 -c anaconda -c conda-forge python=3.6 numpy=1.11.2 pyarrow==0.8.0 pandas nomkl \
+  && $CONDA_BIN create -y -n python2 -c anaconda -c conda-forge python==2.7.15 numpy=1.14.0 pyarrow==0.8.0 pandas nomkl \
+  && $CONDA_BIN create -y -n python3 -c anaconda -c conda-forge python=3.6 numpy=1.14.0 pyarrow==0.8.0 pandas nomkl \
   && $CONDA_BIN clean --all
 
 RUN pyenv global our-miniconda/envs/python2 our-miniconda/envs/python3 \
diff --git a/dev/docker-images/r/Dockerfile b/dev/docker-images/r/Dockerfile
index e660de1de8de4..a4353b3602ddb 100644
--- a/dev/docker-images/r/Dockerfile
+++ b/dev/docker-images/r/Dockerfile
@@ -20,12 +20,8 @@ FROM palantirtechnologies/circle-spark-base
 USER root
 
 ### Install R
-RUN echo "deb http://cran.rstudio.com/bin/linux/ubuntu trusty/" >> /etc/apt/sources.list \
-  && apt-get update \
-  && gpg --keyserver keyserver.ubuntu.com --recv-key E084DAB9 \
-  && gpg -a --export E084DAB9 | sudo apt-key add - \
-  && apt-get update \
-  && apt-get --assume-yes install r-base r-base-dev qpdf \
+RUN apt-get update \
+  && apt-get install r-base r-base-dev qpdf \
   && rm -rf /var/lib/apt/lists/* \
   && chmod 777 /usr/local/lib/R/site-library \
   && /usr/lib/R/bin/R -e "install.packages(c('devtools'), repos='http://cran.us.r-project.org', lib='/usr/local/lib/R/site-library'); devtools::install_github('r-lib/testthat@v2.0.0', lib='/usr/local/lib/R/site-library'); install.packages(c('knitr', 'rmarkdown', 'e1071', 'survival', 'roxygen2', 'lintr'), repos='http://cran.us.r-project.org', lib='/usr/local/lib/R/site-library')"
diff --git a/python/pyspark/tests.py b/python/pyspark/tests.py
index 8397a6eb46ba7..c15d443ebbba9 100644
--- a/python/pyspark/tests.py
+++ b/python/pyspark/tests.py
@@ -2279,20 +2279,20 @@ def test_conda(self):
                     |from pyspark import SparkContext
                     |
                     |sc = SparkContext()
-                    |sc.addCondaPackages('numpy=1.11.1')
+                    |sc.addCondaPackages('numpy=1.14.0')
                     |
                     |# Ensure numpy is accessible on the driver
                     |import numpy
                     |arr = [1, 2, 3]
                     |def mul2(x):
                     |  # Also ensure numpy accessible from executor
-                    |  assert numpy.version.version == "1.11.1"
+                    |  assert numpy.version.version == "1.14.0"
                     |  return x * 2
                     |print(sc.parallelize(arr).map(mul2).collect())
                     """)
         props = self.createTempFile("properties", """
             |spark.conda.binaryPath        {}
-            |spark.conda.channelUrls       https://repo.continuum.io/pkgs/free
+            |spark.conda.channelUrls       https://repo.continuum.io/pkgs/main
             |spark.conda.bootstrapPackages python=3.5
         """.format(os.environ["CONDA_BIN"]))
         env = dict(os.environ)
diff --git a/resource-managers/yarn/src/test/scala/org/apache/spark/deploy/yarn/YarnClusterSuite.scala b/resource-managers/yarn/src/test/scala/org/apache/spark/deploy/yarn/YarnClusterSuite.scala
index f6d3b75a0d4bd..11c06b7f0f906 100644
--- a/resource-managers/yarn/src/test/scala/org/apache/spark/deploy/yarn/YarnClusterSuite.scala
+++ b/resource-managers/yarn/src/test/scala/org/apache/spark/deploy/yarn/YarnClusterSuite.scala
@@ -84,14 +84,14 @@ class YarnClusterSuite extends BaseYarnClusterSuite {
     |        exit(-1)
     |    sc = SparkContext(conf=SparkConf())
     |
-    |    sc.addCondaPackages('numpy=1.11.1')
+    |    sc.addCondaPackages('numpy=1.14.0')
     |    import numpy
     |
     |    status = open(sys.argv[1],'w')
     |
     |    # Addict exists only in external-conda-forge, not anaconda
     |    sc.addCondaChannel("https://conda.anaconda.org/conda-forge")
-    |    sc.addCondaPackages('addict=1.0.0')
+    |    sc.addCondaPackages('addict=2.2.0')
     |
     |    def numpy_multiply(x):
     |        # Ensure package from non-base channel is installed
@@ -376,8 +376,8 @@ class YarnClusterSuite extends BaseYarnClusterSuite {
 
     val extraConf: Map[String, String] = Map(
       "spark.conda.binaryPath" -> sys.env("CONDA_BIN"),
-      "spark.conda.channelUrls" -> "https://repo.continuum.io/pkgs/free",
-      "spark.conda.bootstrapPackages" -> "python=3.5"
+      "spark.conda.channelUrls" -> "https://repo.continuum.io/pkgs/main",
+      "spark.conda.bootstrapPackages" -> "python=3.6"
     )
 
     val moduleDir =