IMPALA-8785: give debug docker images a different name

* Build scripts are generalised to have different targets for release and debug images. * Added new targets for the debug images: docker_debug_images, statestored_debug images. The release images still have the same names. * Separate build contexts are set up for the different base images. * The debug or release base image can be specified as the FROM for the daemon images. * start-impala-cluster.py picks the correct images for the build type Future work: We would like to generalise this to allow building from non-ubuntu-16.04 base images. This probably requires another layer of dockerfiles to specify a base image for impala_base with the required packages installed. Change-Id: I32d2e19cb671beacceebb2642aba01191bd7a244 Reviewed-on: http://gerrit.cloudera.org:8080/13905 Reviewed-by: Joe McDonnell <joemcdonnell@cloudera.com> Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
2025-12-25 02:03:09 -05:00 · 2019-07-23 13:08:35 -07:00
parent dfc968dff1
commit def70c241d
9 changed files with 96 additions and 45 deletions
--- a/bin/jenkins/dockerized-impala-run-tests.sh
+++ b/bin/jenkins/dockerized-impala-run-tests.sh
@@ -72,7 +72,7 @@ start-impala-cluster.py --kill

 # Build the docker images required to start the cluster.
 # parquet-reader is needed for e2e tests but not built for non-test build.
-make -j ${IMPALA_BUILD_THREADS} docker_images parquet-reader
+make -j ${IMPALA_BUILD_THREADS} docker_debug_images parquet-reader

 source_impala_config

--- a/bin/start-impala-cluster.py
+++ b/bin/start-impala-cluster.py
@@ -575,7 +575,12 @@ class DockerMiniClusterOperations(object):
                "-e", "JAVA_TOOL_OPTIONS={0}".format(
                    build_java_tool_options(DEFAULT_IMPALAD_JVM_DEBUG_PORT))]
    # The container build processes tags the generated image with the daemon name.
-    image_tag = daemon
+    debug_build = options.build_type == "debug" or (options.build_type == "latest" and
+        os.path.basename(os.path.dirname(os.readlink("be/build/latest"))) == "debug")
+    if debug_build:
+      image_tag = daemon + "_debug"
+    else:
+      image_tag = daemon
    host_name = self.__gen_host_name__(daemon, instance)
    container_name = self.__gen_container_name__(daemon, instance)
    # Mount configuration into container so that we don't need to rebuild container
--- a/docker/CMakeLists.txt
+++ b/docker/CMakeLists.txt
@@ -19,52 +19,77 @@ set(IMPALA_BASE_BUILD_CONTEXT_DIR
  ${CMAKE_SOURCE_DIR}/docker/build_context
 )

-# Build context depends on daemons and frontend jars
-# Sending the whole impala workspace including test binaries, testdata, etc
-# to the docker daemon can be very expensive, so we create a build context
-# with symlinks
-add_custom_target(impala_base_build_context
-  COMMAND ${CMAKE_SOURCE_DIR}/docker/setup_build_context.py
-  DEPENDS daemons fe ${CMAKE_SOURCE_DIR}/docker/setup_build_context.py
-  COMMENT "Creating impala base build context."
-  VERBATIM
-)
-
-# Target for the base Impala image.
-add_custom_target(impala_base_image
-  # Use tar with -h flag to assemble a tarball including all the symlinked files and
-  # directories in the build context.
-  COMMAND cd ${IMPALA_BASE_BUILD_CONTEXT_DIR} && tar cvh . | docker build -t impala_base -
-  DEPENDS impala_base_build_context ${CMAKE_SOURCE_DIR}/docker/impala_base/Dockerfile
-  DEPENDS ${CMAKE_SOURCE_DIR}/docker/daemon_entrypoint.sh
-  DEPENDS ${CMAKE_SOURCE_DIR}/bin/graceful_shutdown_backends.sh
-  COMMENT "Building Impala base docker image."
-  VERBATIM
-)
+# Add a target to build a base docker image for 'build_type'. 'build_context_args' are
+# passed to the setup_build_context.py script.
+function(add_base_image build_type build_context_args)
+  # Build context depends on daemons and frontend jars
+  # Sending the whole impala workspace including test binaries, testdata, etc
+  # to the docker daemon can be very expensive, so we create a build context
+  # with symlinks
+  add_custom_target(impala_base_build_context_${build_type}
+    COMMAND ${CMAKE_SOURCE_DIR}/docker/setup_build_context.py ${build_context_args}
+    DEPENDS daemons fe ${CMAKE_SOURCE_DIR}/docker/setup_build_context.py
+    COMMENT "Creating impala base build context build_type=${build_type}."
+    VERBATIM
+  )
+  # Target for the base Impala image.
+  add_custom_target(impala_base_image_${build_type}
+    # Use tar with -h flag to assemble a tarball including all the symlinked files and
+    # directories in the build context.
+    COMMAND cd ${IMPALA_BASE_BUILD_CONTEXT_DIR}/${build_type} && tar cvh . | docker build -t impala_base_${build_type} -
+    DEPENDS impala_base_build_context_${build_type} ${CMAKE_SOURCE_DIR}/docker/impala_base/Dockerfile
+    DEPENDS ${CMAKE_SOURCE_DIR}/docker/daemon_entrypoint.sh
+    DEPENDS ${CMAKE_SOURCE_DIR}/bin/graceful_shutdown_backends.sh
+    COMMENT "Building Impala base docker image build_type=${build_type}."
+    VERBATIM
+  )
+endfunction()
+add_base_image(release "")
+add_base_image(debug "--debug-build")

 # Target to build all docker images. Dependencies are added for each docker image
 # instantiated below.
 add_custom_target(docker_images)
+add_custom_target(docker_debug_images)

 set(daemon_image_names "")

-function(add_daemon_docker_image image_name)
-  add_custom_target(${image_name}_image
-    COMMAND cd ${CMAKE_SOURCE_DIR}/docker/${image_name} && docker build -t ${image_name} .
-    DEPENDS impala_base_image ${CMAKE_SOURCE_DIR}/docker/${image_name}/Dockerfile
+# Add a target with name 'target' to build a daemon image for the daemon with
+# name 'daemon_name', e.g. "impalad_executor". The image is tagged as 'image_name'.
+# 'build_type' should be debug or release and determines which base image is used.
+function(add_daemon_docker_image target daemon_name image_name build_type)
+  set(build_dir ${CMAKE_SOURCE_DIR}/docker/${daemon_name})
+  add_custom_target(${target}
+    # Supply the appropriate base image as an argument for the Dockerfile.
+    COMMAND cd ${build_dir} && docker build --build-arg BASE_IMAGE=impala_base_${build_type} -t ${image_name} .
+    DEPENDS impala_base_image_${build_type} ${build_dir}/Dockerfile
    COMMENT "Building ${image_name} docker image."
    VERBATIM
  )
-  ADD_DEPENDENCIES(docker_images ${image_name}_image)
  set(daemon_image_names "${daemon_image_names} ${image_name}" PARENT_SCOPE)
 endfunction()

-# Command and target for the various Impalad images.
-add_daemon_docker_image(impalad_coord_exec)
-add_daemon_docker_image(impalad_coordinator)
-add_daemon_docker_image(impalad_executor)
-add_daemon_docker_image(catalogd)
-add_daemon_docker_image(statestored)
+# Add debug and release docker image targets for the given daemon e.g. if called
+# with "statestored", targets "statestored_image" and "statestored_debug_image"
+# are added.
+function(add_daemon_docker_images daemon_name)
+  set(release_image ${daemon_name})
+  set(release_target ${daemon_name}_image)
+  set(debug_image ${daemon_name}_debug)
+  set(debug_target ${daemon_name}_debug_image)
+  add_daemon_docker_image(${release_target} ${daemon_name} ${release_image} release)
+  add_daemon_docker_image(${debug_target} ${daemon_name} ${debug_image} debug)
+  ADD_DEPENDENCIES(docker_images ${release_target})
+  ADD_DEPENDENCIES(docker_debug_images ${debug_target})
+  set(daemon_image_names "${daemon_image_names} ${release_image}" PARENT_SCOPE)
+endfunction()

-# Generate a test file with all the daemon images.
+# Stamp out image targets for all of the Impala daemons.
+add_daemon_docker_images(impalad_coord_exec)
+add_daemon_docker_images(impalad_coordinator)
+add_daemon_docker_images(impalad_executor)
+add_daemon_docker_images(catalogd)
+add_daemon_docker_images(statestored)
+
+# Generate a text file with all of the release daemon images.
 file(WRITE ${CMAKE_SOURCE_DIR}/docker/docker-images.txt "${daemon_image_names}")
--- a/docker/catalogd/Dockerfile
+++ b/docker/catalogd/Dockerfile
@@ -15,7 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.

-FROM impala_base
+ARG BASE_IMAGE=impala_base
+FROM ${BASE_IMAGE}

 # Externally-facing ports
 # Debug webserver
--- a/docker/impalad_coord_exec/Dockerfile
+++ b/docker/impalad_coord_exec/Dockerfile
@@ -15,7 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.

-FROM impala_base
+ARG BASE_IMAGE=impala_base
+FROM ${BASE_IMAGE}

 # Externally-facing ports
 # Beeswax
--- a/docker/impalad_coordinator/Dockerfile
+++ b/docker/impalad_coordinator/Dockerfile
@@ -15,7 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.

-FROM impala_base
+ARG BASE_IMAGE=impala_base
+FROM ${BASE_IMAGE}

 # Externally-facing ports
 # Beeswax
--- a/docker/impalad_executor/Dockerfile
+++ b/docker/impalad_executor/Dockerfile
@@ -15,7 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.

-FROM impala_base
+ARG BASE_IMAGE=impala_base
+FROM ${BASE_IMAGE}

 # Externally-facing ports
 # Debug webserver
--- a/docker/setup_build_context.py
+++ b/docker/setup_build_context.py
@@ -20,13 +20,23 @@
 # Most artifacts are symlinked so need to be dereferenced (e.g. with tar -h) before
 # being used as a build context.

+import argparse
 import glob
 import os
 import shutil
 from subprocess import check_call

+parser = argparse.ArgumentParser()
+parser.add_argument("--debug-build", help="Setup build context for debug build",
+                    action="store_true")
+args = parser.parse_args()
+
 IMPALA_HOME = os.environ["IMPALA_HOME"]
-OUTPUT_DIR = os.path.join(IMPALA_HOME, "docker/build_context")
+if args.debug_build:
+  BUILD_TYPE = "debug"
+else:
+  BUILD_TYPE = "release"
+OUTPUT_DIR = os.path.join(IMPALA_HOME, "docker/build_context", BUILD_TYPE)
 DOCKERFILE = os.path.join(IMPALA_HOME, "docker/impala_base/Dockerfile")

 IMPALA_TOOLCHAIN = os.environ["IMPALA_TOOLCHAIN"]
@@ -59,9 +69,15 @@ def symlink_file_into_dir(src_file, dst_dir):


 # Impala binaries and native dependencies.
-check_call([STRIP, "--strip-debug",
-            os.path.join(IMPALA_HOME, "be/build/latest/service/impalad"),
-            "-o", os.path.join(BIN_DIR, "impalad")])
+
+# Strip debug symbols from release build to reduce image size. Keep them for
+# debug build.
+IMPALAD_BINARY = os.path.join(IMPALA_HOME, "be/build", BUILD_TYPE, "service/impalad")
+if args.debug_build:
+  symlink_file_into_dir(IMPALAD_BINARY, BIN_DIR)
+else:
+  check_call([STRIP, "--strip-debug", IMPALAD_BINARY,
+              "-o", os.path.join(BIN_DIR, "impalad")])
 for lib in ["libstdc++", "libgcc"]:
  for so in glob.glob(os.path.join(GCC_HOME, "lib64/{0}*.so*".format(lib))):
    symlink_file_into_dir(so, LIB_DIR)
--- a/docker/statestored/Dockerfile
+++ b/docker/statestored/Dockerfile
@@ -15,7 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.

-FROM impala_base
+ARG BASE_IMAGE=impala_base
+FROM ${BASE_IMAGE}

 # Externally-facing ports
 # Debug webserver