Adds clickhouse packaging

.gitignore

@@ -0,0 +1 @@
+*.tar.gz

clickhouse/.gitignore

@@ -0,0 +1 @@
+clickhouse*

clickhouse/README.md

@@ -0,0 +1,41 @@
+# clickhouse
+
+Patches and notes for building ClickHouse on illumos.
+
+## Building
+
+```bash
+$ ./build.sh
+```
+
+## Patches
+
+Most patches have been upstreamed into their various homes. The remaining are
+mostly related to some C++ standard library wierdness. The files in
+`patches/direct` are applied to the source (after cloning submodules), and
+those in `patches/cmake` are applied after running `cmake` as they apply to
+some of the generated build files.
+
+## Upstreaming
+
+In general, ClickHouse was very responsive to PRs, so additional work to
+upstream things should be straightforward. The only bit to record is how to
+handle updates to any of the repos cloned as submodules.
+
+After things have been upstreamed, say to `contrib/project-a`, run
+
+```bash
+$ git submodule update --checkout --remote contrib/project-a
+```
+
+This records the latest commit of the remote submodule into the superproject.
+Once all submodules have been updated like this, make a commit and put up
+a PR against ClickHouse as usual.
+
+## Errors
+
+If you see errors complaining about a submodule not having a particular
+commit or branch, you may need to specify the `branch` in the `.gitmodules`
+file. It should be whatever the remote uses as the default branch. Git
+defaults to using `master`, but many remotes don't use that name, hence
+the errors.

clickhouse/build.sh

@@ -0,0 +1,162 @@
+#!/bin/bash
+
+set -o errexit
+set -o pipefail
+
+function info {
+	printf 'INFO: %s\n' "$*"
+}
+
+function header {
+	printf -- '\n'
+	printf -- '----------------------------------------------------------\n'
+	printf -- 'INFO: %s\n' "$*"
+	printf -- '----------------------------------------------------------\n'
+	printf -- '\n'
+}
+
+function fatal {
+	printf 'ERROR: %s\n' "$*" >&2
+	exit 1
+}
+
+ROOT=$(cd "$(dirname "$0")" && pwd)
+ARTEFACT="$ROOT/clickhouse"
+WORK="$ARTEFACT/build"
+VER="21.7"
+BRANCH="master"
+REPO="https://github.com/oxidecomputer/clickhouse"
+
+# Get platform specific options/tools/paths
+if [ $# -eq 0 ]; then
+    PLATFORM="$OSTYPE"
+else
+    PLATFORM="$1"
+fi
+case $PLATFORM in
+    linux*)
+        PLATFORM="linux"
+        BUILD_COMMAND="make"
+        CC=gcc-10
+        CXX=g++-10
+        STRIP_ARGS="--strip-debug"
+        NPROC="$(nproc)"
+        ;;
+    darwin*)
+        PLATFORM="macos"
+        BUILD_COMMAND="make"
+        CC=clang
+        CXX=clang
+        STRIP_ARGS="-S"
+        NPROC="$(sysctl -n hw.ncpu)"
+        ;;
+    solaris*|illumos*)
+        PLATFORM="illumos"
+        BUILD_COMMAND="ninja"
+        CC=gcc-10
+        CXX=g++-10
+        STRIP_ARGS="-x"
+        NPROC="$(nproc)"
+        ;;
+    *)
+        failed "Unsupported platform $PLATFORM"
+        exit 1
+        ;;
+esac
+COMMON_PATCH_DIR="$ROOT/common/patches"
+PATCH_DIR="$ROOT/$PLATFORM/patches"
+FILES_DIR="$ROOT/$PLATFORM/files"
+EXTRA_FILES=""
+if [ -d "$FILES_DIR" ]; then
+    EXTRA_FILES="$(ls "$FILES_DIR")"
+fi
+header "Building clickhouse for $PLATFORM"
+
+#
+# Download ClickHouse sources
+#
+if [ -d "$ARTEFACT" ]; then
+    info "ClickHouse repo exists, resetting to HEAD"
+    cd "$ARTEFACT"
+    git fetch origin
+    git switch "$BRANCH"
+    git reset --hard "origin/$BRANCH"
+    git submodule update --checkout --recursive --force
+else
+    info "Cloning ClickHouse sources"
+    git clone "$REPO"
+    cd "$ARTEFACT"
+    git switch "$BRANCH"
+    git submodule update --init --recursive
+fi
+
+# Apply common patches, independent of platform
+header "Applying shared ClickHouse patches"
+git apply --verbose $COMMON_PATCH_DIR/*
+
+# Patches to the actual sources. Below we apply those to CMake-generated files.
+if [ -d "$PATCH_DIR/direct" ]; then
+    header "Applying $PLATFORM-specific patches"
+    git apply --verbose $PATCH_DIR/direct/*
+fi
+
+header "Building ClickHouse"
+mkdir -p "$WORK" && cd "$WORK"
+FLAGS="-D_REENTRANT -D_POSIX_PTHREAD_SEMANTICS -D__EXTENSIONS__ -m64 -I$ARTEFACT/contrib/hyperscan-cmake/x86_64/"
+CC=$CC CXX=$CXX CFLAGS="$FLAGS" CXXFLAGS="$FLAGS" \
+cmake \
+    -DABSL_CXX_STANDARD="17" \
+    -DENABLE_LDAP=off \
+    -DUSE_INTERNAL_LDAP_LIBRARY=off \
+    -DENABLE_HDFS=off \
+    -DUSE_INTERNAL_HDFS3_LIBRARY=off \
+    -DENABLE_AMQPCPP=off \
+    -DENABLE_AVRO=off \
+    -DUSE_INTERNAL_AVRO_LIBRARY=off \
+    -DENABLE_CAPNP=off \
+    -DUSE_INTERNAL_CAPNP_LIBRARY=off \
+    -DENABLE_MSGPACK=off \
+    -DUSE_INTERNAL_MSGPACK_LIBRARY=off \
+    -DENABLE_MYSQL=off \
+    -DENABLE_S3=off \
+    -DUSE_INTERNAL_AWS_S3_LIBRARY=off \
+    -DENABLE_PARQUET=off \
+    -DUSE_INTERNAL_PARQUET_LIBRARY=off \
+    -DENABLE_ORC=off \
+    -DUSE_INTERNAL_ORC_LIBRARY=off \
+    -DUSE_SENTRY=off \
+    -DENABLE_CLICKHOUSE_ODBC_BRIDGE=off \
+    -DENABLE_CLICKHOUSE_BENCHMARK=off \
+    -DENABLE_TESTS=off \
+    "$ARTEFACT"
+
+header "Patching CMake-generated files"
+cd "$ARTEFACT"
+if [ -d "$PATCH_DIR/cmake" ]; then
+    git apply --verbose $PATCH_DIR/cmake/*
+fi
+cd "$WORK"
+
+# The build is massive. Try to parallelize until we error out, usually due to space constraints while
+# linking. At that point, continue serially
+$BUILD_COMMAND -j "$NPROC" || (header "Parallel build failed, continuing serially" && $BUILD_COMMAND -j 1)
+
+# Strip the resulting binary. This part is crucial. ClickHouse's binary is 3+GiB unstripped.
+strip $STRIP_ARGS "$WORK/programs/clickhouse"
+CONFIG_FILE_DIR="$ARTEFACT/programs/server"
+CONFIG_FILE_NAME="config.xml"
+if [ -z "$EXTRA_FILES" ]; then
+    /usr/bin/tar cvfz \
+        $ROOT/clickhouse-v$VER.$PLATFORM.tar.gz \
+        -C "$WORK/programs" clickhouse \
+        -C "$CONFIG_FILE_DIR" "$CONFIG_FILE_NAME"
+else
+    /usr/bin/tar cvfz \
+        $ROOT/clickhouse-v$VER.$PLATFORM.tar.gz \
+        -C "$WORK/programs" clickhouse \
+        -C "$CONFIG_FILE_DIR" "$CONFIG_FILE_NAME" \
+        -C "$FILES_DIR" "$EXTRA_FILES"
+fi
+
+header "Build output:"
+find "$WORK" -type f -ls

clickhouse/common/patches/config.patch

@@ -0,0 +1,64 @@
+diff --git a/programs/server/config.xml b/programs/server/config.xml
+index df8a5266c3..e58318e970 100644
+--- a/programs/server/config.xml
++++ b/programs/server/config.xml
+@@ -22,8 +22,8 @@
+             [1]: https://github.com/pocoproject/poco/blob/poco-1.9.4-release/Foundation/include/Poco/Logger.h#L105-L114
+         -->
+         <level>trace</level>
+-        <log>/var/log/clickhouse-server/clickhouse-server.log</log>
+-        <errorlog>/var/log/clickhouse-server/clickhouse-server.err.log</errorlog>
++        <log>/tmp/clickhouse/21.7/clickhouse-server.log</log>
++        <errorlog>/tmp/clickhouse/21.7/clickhouse-server.err.log</errorlog>
+         <!-- Rotation policy
+              See https://github.com/pocoproject/poco/blob/poco-1.9.4-release/Foundation/include/Poco/FileChannel.h#L54-L85
+           -->
+@@ -333,10 +333,10 @@
+     <compiled_expression_cache_size>1073741824</compiled_expression_cache_size>
+
+     <!-- Path to data directory, with trailing slash. -->
+-    <path>/var/lib/clickhouse/</path>
++    <path>/tmp/clickhouse/21.7/</path>
+
+     <!-- Path to temporary data for processing hard queries. -->
+-    <tmp_path>/var/lib/clickhouse/tmp/</tmp_path>
++    <tmp_path>/tmp/clickhouse/21.7/tmp/</tmp_path>
+
+     <!-- Policy from the <storage_configuration> for the temporary files.
+          If not set <tmp_path> is used, otherwise <tmp_path> is ignored.
+@@ -350,7 +350,7 @@
+     <!-- <tmp_policy>tmp</tmp_policy> -->
+
+     <!-- Directory with user provided files that are accessible by 'file' table function. -->
+-    <user_files_path>/var/lib/clickhouse/user_files/</user_files_path>
++    <user_files_path>/tmp/clickhouse/21.7/user_files/</user_files_path>
+
+     <!-- LDAP server definitions. -->
+     <ldap_servers>
+@@ -425,7 +425,7 @@
+         </users_xml>
+         <local_directory>
+             <!-- Path to folder where users created by SQL commands are stored. -->
+-            <path>/var/lib/clickhouse/access/</path>
++            <path>/tmp/clickhouse/21.7/access/</path>
+         </local_directory>
+
+         <!-- To add an LDAP server as a remote user directory of users that are not defined locally, define a single 'ldap' section
+@@ -927,7 +927,7 @@
+     <!-- <path_to_regions_names_files>/opt/geo/</path_to_regions_names_files> -->
+
+
+-    <!-- <top_level_domains_path>/var/lib/clickhouse/top_level_domains/</top_level_domains_path> -->
++    <!-- <top_level_domains_path>/tmp/clickhouse/21.7/top_level_domains/</top_level_domains_path> -->
+     <!-- Custom TLD lists.
+          Format: <name>/path/to/file</name>
+
+@@ -1040,7 +1040,7 @@
+     <!-- Directory in <clickhouse-path> containing schema files for various input formats.
+          The directory will be created if it doesn't exist.
+       -->
+-    <format_schema_path>/var/lib/clickhouse/format_schemas/</format_schema_path>
++    <format_schema_path>/tmp/clickhouse/21.7/format_schemas/</format_schema_path>
+
+     <!-- Default query masking rules, matching lines would be replaced with something else in the logs
+         (both text logs and system.query_log).

clickhouse/common/patches/embedded.patch

@@ -0,0 +1,13 @@
+diff --git a/programs/server/embedded.xml b/programs/server/embedded.xml
+index a66f57d1eb..bfb374091c 100644
+--- a/programs/server/embedded.xml
++++ b/programs/server/embedded.xml
+@@ -10,7 +10,7 @@
+     <tcp_port>9000</tcp_port>
+     <mysql_port>9004</mysql_port>
+
+-    <path>./</path>
++    <path>/tmp</path>
+
+     <uncompressed_cache_size>8589934592</uncompressed_cache_size>
+     <mark_cache_size>5368709120</mark_cache_size>

clickhouse/illumos/files/manifest.xml

@@ -0,0 +1,35 @@
+<?xml version="1.0"?>
+<!DOCTYPE service_bundle SYSTEM "/usr/share/lib/xml/dtd/service_bundle.dtd.1">
+
+<service_bundle type='manifest' name='clickhouse'>
+
+<service name='system/illumos/clickhouse' type='service' version='1'>
+  <create_default_instance enabled='false' />
+  <single_instance />
+
+  <exec_method type='method' name='start'
+    exec='ctrun -l child -o noorphan,regent /opt/clickhouse/21.7/clickhouse server --config-file %{config/config_file} &amp;'
+    timeout_seconds='0' />
+  <exec_method type='method' name='stop' exec=':kill' timeout_seconds='0' />
+
+  <property_group name='config' type='application'>
+    <propval name='config_file' type='astring' value='/opt/clickhouse/21.7/config.xml' />
+  </property_group>
+
+  <property_group name='startd' type='framework'>
+    <propval name='duration' type='astring' value='contract' />
+  </property_group>
+
+  <stability value='Unstable' />
+
+  <template>
+    <common_name>
+      <loctext xml:lang='C'>ClickHouse</loctext>
+    </common_name>
+    <description>
+      <loctext xml:lang='C'>Oxide metric database</loctext>
+    </description>
+  </template>
+</service>
+
+</service_bundle>

clickhouse/illumos/patches/cmake/boringssl-cmake.patch

@@ -0,0 +1,18 @@
+diff --git a/contrib/boringssl-cmake/CMakeLists.txt b/contrib/boringssl-cmake/CMakeLists.txt
+index 017a8a6..7e68e7c 100644
+--- a/contrib/boringssl-cmake/CMakeLists.txt
++++ b/contrib/boringssl-cmake/CMakeLists.txt
+@@ -78,7 +78,11 @@ elseif(${CMAKE_SYSTEM_PROCESSOR} STREQUAL "AMD64")
+ elseif(${CMAKE_SYSTEM_PROCESSOR} STREQUAL "x86")
+   set(ARCH "x86")
+ elseif(${CMAKE_SYSTEM_PROCESSOR} STREQUAL "i386")
+-  set(ARCH "x86")
++  if(${CMAKE_SYSTEM_NAME} STREQUAL "SunOS")
++    set(ARCH "x86_64")
++  else()
++    set(ARCH "x86")
++  endif()
+ elseif(${CMAKE_SYSTEM_PROCESSOR} STREQUAL "i686")
+   set(ARCH "x86")
+ elseif(${CMAKE_SYSTEM_PROCESSOR} STREQUAL "aarch64")
+Submodule contrib/cppkafka contains modified content

clickhouse/illumos/patches/cmake/poco-cmake.patch

@@ -0,0 +1,16 @@
+diff --git a/contrib/poco-cmake/Net/CMakeLists.txt b/contrib/poco-cmake/Net/CMakeLists.txt
+index 9bc06e5..dd6894f 100644
+--- a/contrib/poco-cmake/Net/CMakeLists.txt
++++ b/contrib/poco-cmake/Net/CMakeLists.txt
+@@ -127,6 +127,10 @@ if (USE_INTERNAL_POCO_LIBRARY)
+     )
+     target_include_directories (_poco_net SYSTEM PUBLIC "${LIBRARY_DIR}/Net/include")
+     target_link_libraries (_poco_net PUBLIC Poco::Foundation)
++
++    if (OS_SUNOS)
++		target_link_libraries (_poco_net PUBLIC socket nsl)
++    endif()
+ else ()
+     add_library (Poco::Net UNKNOWN IMPORTED GLOBAL)
+
+Submodule contrib/rocksdb contains modified content

clickhouse/illumos/patches/direct/AbseilConfigureCopts.patch

@@ -0,0 +1,10 @@
+diff --git a/contrib/abseil-cpp/absl/copts/AbseilConfigureCopts.cmake b/contrib/abseil-cpp/absl/copts/AbseilConfigureCopts.cmake
+index 77d4ace..30891a4 100644
+--- a/contrib/abseil-cpp/absl/copts/AbseilConfigureCopts.cmake
++++ b/contrib/abseil-cpp/absl/copts/AbseilConfigureCopts.cmake
+@@ -72,3 +72,5 @@ elseif(NOT "${CMAKE_CXX_STANDARD}")
+ else()
+   set(ABSL_CXX_STANDARD "${CMAKE_CXX_STANDARD}")
+ endif()
++# Absl cannot work with C++20, which removed some typedefs for std::allocator
++set(ABSL_CXX_STANDARD 17)

clickhouse/illumos/patches/direct/PerformanceAdaptors.patch

@@ -0,0 +1,13 @@
+diff --git a/src/Functions/PerformanceAdaptors.h b/src/Functions/PerformanceAdaptors.h
+index 5d4d013..716bb65 100644
+--- a/src/Functions/PerformanceAdaptors.h
++++ b/src/Functions/PerformanceAdaptors.h
+@@ -99,7 +99,7 @@ namespace detail
+             ///  when there is no statistical significant difference between them.
+             double sigma() const
+             {
+-                return mean() / sqrt(adjustedCount());
++                return mean() / sqrt(static_cast<double>(adjustedCount()));
+             }
+
+             void run()