Skip to content

Commit

Permalink
[fix] Switch docker base image to avoid mixing musl & glibc libraries…
Browse files Browse the repository at this point in the history
… at runtime

Mixing musl & glibc libraries at runtime will result in compatibility issues.
We need a solution with real glibc because of compatibility reasons.
The previous solution with the Alpine base image has a critical issue since
it mixes musl compiled and glibc compiled libraries at runtime while using real glibc.
This is why the switch is needed.

For example, snappy-java and Conscrypt show that this problem occurs.
  • Loading branch information
lhotari committed Sep 28, 2024
1 parent 95bd1d1 commit 970c08b
Show file tree
Hide file tree
Showing 6 changed files with 14 additions and 268 deletions.
80 changes: 0 additions & 80 deletions docker/glibc-package/Dockerfile

This file was deleted.

39 changes: 0 additions & 39 deletions docker/glibc-package/README.md

This file was deleted.

53 changes: 0 additions & 53 deletions docker/glibc-package/scripts/APKBUILD

This file was deleted.

21 changes: 0 additions & 21 deletions docker/glibc-package/scripts/glibc-bin.trigger

This file was deleted.

23 changes: 0 additions & 23 deletions docker/glibc-package/scripts/ld.so.conf

This file was deleted.

66 changes: 14 additions & 52 deletions docker/pulsar/Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -17,6 +17,9 @@
# under the License.
#

# Final image uses the Liberica JDK image running on Alpaquita Linux as the base image
ARG LIBERICA_RUNTIME_CONTAINER_TAG=jdk-all-17-cds-stream-glibc
# Alpine is used for intermediate stages
ARG ALPINE_VERSION=3.20

# First create a stage with just the Pulsar tarball and scripts
Expand Down Expand Up @@ -50,55 +53,29 @@ RUN for SUBDIRECTORY in conf data download logs instances/deps packages-storage;
RUN chmod -R g+rx /pulsar/bin
RUN chmod -R o+rx /pulsar

# Enable snappy-java to use system lib
RUN echo 'OPTS="$OPTS -Dorg.xerial.snappy.use.systemlib=true"' >> /pulsar/conf/bkenv.sh

### Create one stage to include JVM distribution
FROM amazoncorretto:21-alpine AS jvm

RUN apk add --no-cache binutils

# Use JLink to create a slimmer JDK distribution (see: https://adoptium.net/blog/2021/10/jlink-to-produce-own-runtime/)
# This still includes all JDK modules, though in the future we could compile a list of required modules
RUN /usr/lib/jvm/default-jvm/bin/jlink --add-modules ALL-MODULE-PATH --compress zip-9 --no-man-pages --no-header-files --strip-debug --output /opt/jvm
RUN echo networkaddress.cache.ttl=1 >> /opt/jvm/conf/security/java.security
RUN echo networkaddress.cache.negative.ttl=1 >> /opt/jvm/conf/security/java.security

## Create one stage to include snappy-java native lib
# Fix the issue when using snappy-java in x86 arch alpine
# See https://github.com/xerial/snappy-java/issues/181 https://github.com/xerial/snappy-java/issues/579
# We need to ensure that the version of the native library matches the version of snappy-java imported via Maven
FROM alpine:$ALPINE_VERSION AS snappy-java

ARG SNAPPY_VERSION
RUN apk add git alpine-sdk util-linux cmake autoconf automake libtool openjdk17 maven curl bash tar
ENV JAVA_HOME=/usr
RUN curl -Ls https://github.com/xerial/snappy-java/archive/refs/tags/v$SNAPPY_VERSION.tar.gz | tar zxf - && cd snappy-java-$SNAPPY_VERSION && make clean-native native
FROM apachepulsar/glibc-base:2.38 as glibc

## Create final stage from Alpine image
## and add OpenJDK and Python dependencies (for Pulsar functions)
FROM alpine:$ALPINE_VERSION
## Create final stage from liberica-runtime-container image
## and add Python dependencies (for Pulsar functions)
FROM bellsoft/liberica-runtime-container:$LIBERICA_RUNTIME_CONTAINER_TAG
ENV LANG C.UTF-8

# Upgrade all packages to get latest versions with security fixes
# Install some utilities, some are required by Pulsar scripts
RUN apk add --no-cache \
RUN apk upgrade --no-cache && apk add --no-cache \
bash \
python3 \
py3-pip \
py3-grpcio \
py3-yaml \
gcompat \
ca-certificates \
procps \
curl \
bind-tools
bind-tools \
openssl

# Upgrade all packages to get latest versions with security fixes
RUN apk upgrade --no-cache
# adjust default DNS caching TTL to 1 second
RUN echo networkaddress.cache.ttl=1 >> $JAVA_HOME/conf/security/java.security && echo networkaddress.cache.negative.ttl=1 >> $JAVA_HOME/conf/security/java.security

# Python dependencies

# The [email protected] is installed by apk, and [email protected] requires grpcio>=1.60.0, which causes the grocio to be reinstalled by pip.
# If pip cannot find the grpcio wheel that the doesn't match the OS, the grpcio will be compiled locally.
# Once https://github.com/apache/pulsar-client-python/pull/211 is released, keep only the pulsar-client[all] and kazoo dependencies, and remove comments.
Expand All @@ -116,31 +93,16 @@ prometheus_client\n\
ratelimit\n\
# avro\n\
fastavro>=1.9.2\n\
" > /requirements.txt

RUN pip3 install --break-system-packages --no-cache-dir --only-binary grpcio -r /requirements.txt
RUN rm /requirements.txt

# Install GLibc compatibility library
COPY --from=glibc /root/packages /root/packages
RUN apk add --allow-untrusted --force-overwrite /root/packages/glibc-*.apk

COPY --from=jvm /opt/jvm /opt/jvm
ENV JAVA_HOME=/opt/jvm

COPY --from=snappy-java /tmp/libsnappyjava.so /usr/lib/libsnappyjava.so
" > /tmp/requirements.txt && RUN pip3 install --break-system-packages --no-cache-dir --only-binary grpcio -r /tmp/requirements.txt && rm /tmp/requirements.txt

# The default is /pulsat/bin and cannot be written.
ENV PULSAR_PID_DIR=/pulsar/logs

ENV PULSAR_ROOT_LOGGER=INFO,CONSOLE

COPY --from=pulsar /pulsar /pulsar

WORKDIR /pulsar
ENV PATH=$PATH:$JAVA_HOME/bin:/pulsar/bin

# The UID must be non-zero. Otherwise, it is arbitrary. No logic should rely on its specific value.
ARG DEFAULT_USERNAME=pulsar
RUN adduser ${DEFAULT_USERNAME} -u 10000 -G root -D -H -h /pulsar/data
USER 10000
USER 10000

0 comments on commit 970c08b

Please sign in to comment.