--- title: Tanzu Greenplum 7.8をOrbStack上のRocky Linux Machineにインストールするメモ summary: OrbStackでApple Silicon Mac上にGreenplumクラスタを構築する方法を紹介します。Cloud Initで自動設定し、動作確認まで行います。 tags: ["Greenplum", "Tanzu", "PostgreSQL", "Rocky", "OrbStack"] categories: ["Middleware", "RDBMS", "Greenplum"] date: 2026-05-22T04:50:55.866Z updated: 2026-05-22T04:52:37.542Z --- [以前の記事](/entries/879)でIntel MacのVMware Fusion上にGreenplumを構築しましたが、やっぱりApple SiliconのMac上で環境を作りたかったので、今回は[OrbStack](https://orbstack.dev/)の[Linux machines](https://docs.orbstack.dev/machines/)上で構築します。x86_64はRosettaによるエミュレートなので遅いですが、手軽さを優先しました。 ### Tanzu Greenplumのダウンロード GreenplumのRPMパッケージをダウンロードします。Broadcomのアカウントが必要です。 ホストのMac上で`~/greenplum`ディレクトリを作成します。 ```bash mkdir -p ~/greenplum ``` [Broadcom Support](https://support.broadcom.com)にログインして、[Tanzu Greenplumのダウンロードページ](https://support.broadcom.com/group/ecx/productdownloads?subfamily=VMware%20Tanzu%20Greenplum)にアクセスします。 最新バージョンを選択します。 ![image](https://s3.ik.am/ikam/_/1779414669094_pasted-image.png) RHEL 9用のRPMパッケージをダウンロードします。必要に応じて他のパッケージもダウンロードしてください。 ![image](https://s3.ik.am/ikam/_/1779414727188_pasted-image.png) ## OrbStackでGreenplumクラスタの作成 Linux MachineのCloud Initで一気にGreenplumクラスタを作成します。 次の`user-data.yml`を作成します。 IPアドレスの`192.168.139`の部分や、rpmのパス`GP_RPM`は環境に合わせて変更してください。 ```yaml #cloud-config # ============================================================================ # Greenplum 7 cluster bootstrap — single-file cloud-config # # This file drives the entire procedure WITHOUT Ansible: # - OS prep (packages, sysctl, limits, firewall, SELinux) # - gpadmin user, sudoers, .bashrc # - Static network configuration # - Greenplum RPM download + install # - /data directories per role # - SSH key exchange between all nodes (gpssh-exkeys) # - gpinitsystem (coordinator only) # # The role (coordinator / segment) is auto-detected from the hostname, # so the same file works for all 4 nodes. # # # Topology (default): # gp-coordinator 192.168.139.200/24 # gp-segment1 192.168.139.201/24 # gp-segment2 192.168.139.202/24 # gp-segment3 192.168.139.203/24 # # Boot order: # Start all 4 VMs at the same time. The coordinator's runcmd waits up to # 30 minutes per segment for SSH to come up before continuing. # ============================================================================ manage_etc_hosts: false # we ship /etc/hosts via write_files # ---------- SSH ------------------------------------------------------------- # Password auth must be enabled so that gpssh-exkeys / ssh-copy-id work # during initial key exchange. Tighten afterwards if desired. ssh_pwauth: true disable_root: false users: - name: root lock_passwd: false ssh_authorized_keys: # (*) your public key - ssh-rsa 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 - ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIP1p/lDTw/mPvjTq1BVV6htcEQa+xSg+hG9tM0Ky9jYw - name: gpadmin primary_group: gpadmin shell: /bin/bash lock_passwd: false sudo: "ALL=(ALL) NOPASSWD:ALL" groups: - gpadmin chpasswd: expire: false users: - {name: root, password: rocky, type: text} - {name: gpadmin, password: Greenplum123, type: text} # ---------- Packages -------------------------------------------------------- package_update: true packages: - apr - apr-util - bzip2 - curl - iproute - krb5-devel - libcurl - libevent - libtiff - libuuid - libuv - libxml2 - libyaml - libzstd - lsof - openldap - openssh - openssh-clients - openssh-server - openssl - openssl-libs - perl - python3 - python3-psycopg2 - python3-psutil - python3-pyyaml - python3.11 - python3.11-devel - readline - rsync - sed - tar - which - zip - zlib - sshpass - java-21-openjdk-devel - epel-release # ---------- Static files ---------------------------------------------------- write_files: - path: /etc/hosts permissions: '0644' content: | 127.0.0.1 localhost 192.168.139.200 gp-coordinator 192.168.139.201 gp-segment1 192.168.139.202 gp-segment2 192.168.139.203 gp-segment3 - path: /etc/sysctl.d/99-greenplum.conf content: | net.ipv4.ip_local_reserved_ports=65330 # file descriptor limits - path: /etc/security/limits.d/99-greenplum.conf content: | * soft nofile 65536 * hard nofile 65536 # ----- Static network configuration ------------------------------------- - path: /usr/local/sbin/cloud-setup-network.sh permissions: '0755' owner: root:root content: | #!/bin/bash set -eux # Derive last octet from hostname: gp-coordinator -> 200, gp-segmentN -> 20N HOST=$(hostname -s) if [[ "$HOST" == "gp-coordinator" ]]; then OCTET=200 elif [[ "$HOST" =~ ^gp-segment([0-9]+)$ ]]; then OCTET=$((200 + BASH_REMATCH[1])) else echo "unknown host: $HOST" >&2; exit 1 fi IPADDR="192.168.139.${OCTET}/24" FALLBACK_GW="192.168.139.1" DNS="8.8.8.8,1.1.1.1" NIC=$(ip -o -4 route show default 2>/dev/null | awk '{print $5}' | head -n1) GATEWAY=$(ip -o -4 route show default 2>/dev/null | awk '{print $3}' | head -n1) [ -z "$NIC" ] && NIC=$(ls /sys/class/net | grep -E '^(en|eth)' | grep -v '^lo' | head -n1) [ -z "$NIC" ] && { echo "no NIC found"; exit 1; } [ -z "$GATEWAY" ] && GATEWAY="$FALLBACK_GW" for con in $(nmcli -t -f NAME,DEVICE con show | awk -F: -v n="$NIC" '$2==n{print $1}'); do nmcli con delete "$con" || true done nmcli con add type ethernet ifname "$NIC" con-name static \ ipv4.method manual ipv4.addresses "$IPADDR" \ ipv4.gateway "$GATEWAY" ipv4.dns "$DNS" \ ipv6.method disabled connection.autoconnect yes nmcli con up static # ----- Greenplum node provisioner (full procedure) ---------------------- # Determines role from hostname, installs Greenplum, prepares /data, # and (on coordinator only) runs the cluster init. - path: /usr/local/sbin/cloud-provision-greenplum.sh permissions: '0755' owner: root:root content: | #!/bin/bash set -euxo pipefail exec > >(tee -a /var/log/cloud-provision-greenplum.log) 2>&1 ulimit -SHn 65536 || true echo "[provision] ulimit -n = $(ulimit -n) (hard: $(ulimit -Hn))" # ===== Settings (edit if needed) =================================== GP_RPM=/Users/toshiaki/greenplum/greenplum-db-7.8.1-el9-x86_64.rpm GPADMIN_PW="Greenplum123" COORDINATOR_HOST="gp-coordinator" ALL_HOSTS=(gp-coordinator gp-segment1 gp-segment2 gp-segment3) SEGMENT_HOSTS=(gp-segment1 gp-segment2 gp-segment3) # =================================================================== HOSTNAME_SHORT=$(hostname -s) if [[ "$HOSTNAME_SHORT" == "$COORDINATOR_HOST" ]]; then ROLE=coordinator else ROLE=segment fi echo "[provision] role=$ROLE host=$HOSTNAME_SHORT" # ---- Java default --------------------------------------------------- JAVA21=$(ls /usr/lib/jvm/java-21-openjdk-*/bin/java 2>/dev/null | head -n1 || true) [ -n "$JAVA21" ] && alternatives --set java "$JAVA21" || true # ---- firewalld / SELinux ------------------------------------------- systemctl disable --now firewalld 2>/dev/null || true setenforce 0 2>/dev/null || true sed -i 's/^SELINUX=enforcing/SELINUX=disabled/' /etc/selinux/config || true # ---- Download & install Greenplum RPM ------------------------------ dnf install -y $GP_RPM GPHOME=$(ls -d /usr/local/greenplum-db-* 2>/dev/null | sort | tail -n1) chown -R gpadmin:gpadmin "$GPHOME" # ---- gpadmin .bashrc ----------------------------------------------- su - gpadmin -c " grep -q greenplum_path ~/.bashrc \ || echo 'source $GPHOME/greenplum_path.sh' >> ~/.bashrc " # ---- /data directories --------------------------------------------- if [[ "$ROLE" == "coordinator" ]]; then mkdir -p /data/coordinator else mkdir -p /data/primary /data/mirror fi chown -R gpadmin:gpadmin /data # ---- gpadmin SSH keypair ------------------------------------------- su - gpadmin -c ' [ -f ~/.ssh/id_rsa ] || ssh-keygen -m PEM -t rsa -b 4096 -q -N "" -f ~/.ssh/id_rsa ' # ---- Coordinator-only: cluster init -------------------------------- if [[ "$ROLE" == "coordinator" ]]; then echo "[provision] coordinator: waiting for all nodes to accept SSH..." for h in "${ALL_HOSTS[@]}"; do for i in $(seq 1 30); do if sshpass -p "$GPADMIN_PW" ssh \ -o StrictHostKeyChecking=no \ -o ConnectTimeout=2 \ -o PasswordAuthentication=yes \ -o PreferredAuthentications=password \ gpadmin@"$h" true 2>/dev/null; then echo "[provision] $h is up" break fi [[ $i -eq 30 ]] && { echo "[provision] timeout waiting for $h"; exit 1; } sleep 5 done done # SSH key exchange (per the original procedure) for h in "${ALL_HOSTS[@]}"; do su - gpadmin -c " SSHPASS='$GPADMIN_PW' sshpass -e ssh-copy-id -o StrictHostKeyChecking=no $h " done # gpssh-exkeys populates known_hosts cluster-wide su - gpadmin -c " printf '%s\n' ${ALL_HOSTS[*]} > ~/hostfile_exkeys source $GPHOME/greenplum_path.sh gpssh-exkeys -f ~/hostfile_exkeys " # gpinitsystem_config su - gpadmin -c " mkdir -p ~/gpconfigs printf '%s\n' ${SEGMENT_HOSTS[*]} > ~/gpconfigs/hostfile_gpinitsystem cp $GPHOME/docs/cli_help/gpconfigs/gpinitsystem_config ~/gpconfigs/ sed -i \ -e 's|^declare -a DATA_DIRECTORY=.*|declare -a DATA_DIRECTORY=(/data/primary /data/primary)|' \ -e 's|^#declare -a MIRROR_DATA_DIRECTORY=.*|declare -a MIRROR_DATA_DIRECTORY=(/data/mirror /data/mirror)|' \ -e 's|^#MIRROR_PORT_BASE=.*|MIRROR_PORT_BASE=7000|' \ -e 's|^COORDINATOR_HOSTNAME=.*|COORDINATOR_HOSTNAME=$COORDINATOR_HOST|' \ ~/gpconfigs/gpinitsystem_config " # Initialize the cluster (non-interactive) su - gpadmin -c " ulimit -n 65536 source $GPHOME/greenplum_path.sh gpinitsystem -a \ -c ~/gpconfigs/gpinitsystem_config \ -h ~/gpconfigs/hostfile_gpinitsystem " # Greenplum env vars for gpadmin printf '%s\n' \ 'export COORDINATOR_DATA_DIRECTORY=/data/coordinator/gpseg-1' \ 'export PGPORT=5432' \ 'export PGUSER=gpadmin' \ 'export PGDATABASE=gpadmin' \ >> /home/gpadmin/.bashrc chown gpadmin:gpadmin /home/gpadmin/.bashrc fi echo "[provision] done role=$ROLE" # ---------- Boot sequence --------------------------------------------------- runcmd: # 0. Make sure sshd is running and enabled (minimal/container images often # ship with it disabled, which would block gpssh-exkeys / ssh-copy-id) - [systemctl, enable, --now, sshd] # 1. Switch from DHCP to static IP - /usr/local/sbin/cloud-setup-network.sh # 2. Apply sysctl (limits picks up on next login automatically) - [sysctl, --system] # 3. Full Greenplum provisioning (coordinator also runs gpinitsystem) - /usr/local/sbin/cloud-provision-greenplum.sh # ---------- Final message --------------------------------------------------- final_message: | cloud-init finished in $UPTIME seconds. See /var/log/cloud-provision-greenplum.log for Greenplum bootstrap output. If this is the coordinator, run as gpadmin: gpstate -s # cluster status gpstate -e # mirror status ``` この`user-data.yml`を使ってCoordinatorとSegmentを一気に作ります。 ```bash orb create -a amd64 rocky:9 gp-coordinator -c user-data.yml orb create -a amd64 rocky:9 gp-segment1 -c user-data.yml orb create -a amd64 rocky:9 gp-segment2 -c user-data.yml orb create -a amd64 rocky:9 gp-segment3 -c user-data.yml ``` ### 構成の確認 Coordinatorにログインします。 ```bash orb shell -m gp-coordinator -u gpadmin -w /home/gpadmin ``` ```bash # クラスタ状態確認 gpstate -s # セグメント詳細確認 gpstate -e # ミラー状態確認 gpstate -m ``` セグメント詳細確認は次のように表示されます。 ```bash $ gpstate -e 20260522:13:19:42:014744 gpstate:gp-coordinator:gpadmin-[INFO]:-Starting gpstate with args: -e 20260522:13:19:42:014744 gpstate:gp-coordinator:gpadmin-[INFO]:-local Greenplum Version: 'postgres (Greenplum Database) 7.8.1 build commit:2ac9af769d652200c3a913cc8ae39d54d7d72ded' 20260522:13:19:42:014744 gpstate:gp-coordinator:gpadmin-[INFO]:-Coordinator Greenplum Version: 'PostgreSQL 12.22 (Greenplum Database 7.8.1 build commit:2ac9af769d652200c3a913cc8ae39d54d7d72ded) on x86_64-pc-linux-gnu, compiled by gcc (GCC) 11.5.0 20240719 (Red Hat 11.5.0-11), 64-bit compiled on May 15 2026 07:32:19 Bhuvnesh C.' 20260522:13:19:42:014744 gpstate:gp-coordinator:gpadmin-[INFO]:-Obtaining Segment details from coordinator... 20260522:13:19:42:014744 gpstate:gp-coordinator:gpadmin-[INFO]:-Gathering data from segments... ..... 20260522:13:19:51:014744 gpstate:gp-coordinator:gpadmin-[INFO]:----------------------------------------------------- 20260522:13:19:51:014744 gpstate:gp-coordinator:gpadmin-[INFO]:-Segment Mirroring Status Report 20260522:13:19:54:014744 gpstate:gp-coordinator:gpadmin-[INFO]:----------------------------------------------------- 20260522:13:19:54:014744 gpstate:gp-coordinator:gpadmin-[INFO]:-All segments are running normally ``` ミラー状態確認は次のように表示されます。 ```bash $ gpstate -m 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:-Starting gpstate with args: -m 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:-local Greenplum Version: 'postgres (Greenplum Database) 7.8.1 build commit:2ac9af769d652200c3a913cc8ae39d54d7d72ded' 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:-Coordinator Greenplum Version: 'PostgreSQL 12.22 (Greenplum Database 7.8.1 build commit:2ac9af769d652200c3a913cc8ae39d54d7d72ded) on x86_64-pc-linux-gnu, compiled by gcc (GCC) 11.5.0 20240719 (Red Hat 11.5.0-11), 64-bit compiled on May 15 2026 07:32:19 Bhuvnesh C.' 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:-Obtaining Segment details from coordinator... 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:-------------------------------------------------------------- 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:--Current GPDB mirror list and status 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:--Type = Group 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:-------------------------------------------------------------- 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:- Mirror Datadir Port Status Data Status 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:- gp-segment2 /data/mirror/gpseg0 7000 Passive Synchronized 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:- gp-segment2 /data/mirror/gpseg1 7001 Passive Synchronized 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:- gp-segment3 /data/mirror/gpseg2 7000 Passive Synchronized 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:- gp-segment3 /data/mirror/gpseg3 7001 Passive Synchronized 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:- gp-segment1 /data/mirror/gpseg4 7000 Passive Synchronized 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:- gp-segment1 /data/mirror/gpseg5 7001 Passive Synchronized 20260522:13:20:16:014786 gpstate:gp-coordinator:gpadmin-[INFO]:-------------------------------------------------------------- ``` `psql`でも構成を確認します。 ```bash psql -d postgres -c "SELECT content, role, preferred_role, hostname, port, datadir, status FROM gp_segment_configuration ORDER BY content, role;" ``` 以下の構成になっていることを確認します。 ``` content | role | preferred_role | hostname | port | datadir | status ---------+------+----------------+----------------+------+---------------------------+-------- -1 | p | p | gp-coordinator | 5432 | /data/coordinator/gpseg-1 | u 0 | m | m | gp-segment2 | 7000 | /data/mirror/gpseg0 | u 0 | p | p | gp-segment1 | 6000 | /data/primary/gpseg0 | u 1 | m | m | gp-segment2 | 7001 | /data/mirror/gpseg1 | u 1 | p | p | gp-segment1 | 6001 | /data/primary/gpseg1 | u 2 | m | m | gp-segment3 | 7000 | /data/mirror/gpseg2 | u 2 | p | p | gp-segment2 | 6000 | /data/primary/gpseg2 | u 3 | m | m | gp-segment3 | 7001 | /data/mirror/gpseg3 | u 3 | p | p | gp-segment2 | 6001 | /data/primary/gpseg3 | u 4 | m | m | gp-segment1 | 7000 | /data/mirror/gpseg4 | u 4 | p | p | gp-segment3 | 6000 | /data/primary/gpseg4 | u 5 | m | m | gp-segment1 | 7001 | /data/mirror/gpseg5 | u 5 | p | p | gp-segment3 | 6001 | /data/primary/gpseg5 | u (13 rows) ``` **構成のポイント:** - 合計6つのプライマリセグメント（content 0-5） - 各プライマリに対応するミラーが別ホストに配置 - 各サーバーに2プライマリ + 2ミラー = 計4インスタンス ### 動作確認 テストデータベースを作成します。 ```bash createdb test ``` テーブルを作成し、動作確認します。 ```bash psql -d test -c "CREATE TABLE test_table (id int, name text) DISTRIBUTED BY (id);" psql -d test -c "INSERT INTO test_table SELECT generate_series(1,1000), 'test_data';" psql -d test -c "SELECT gp_segment_id, count(*) FROM test_table GROUP BY gp_segment_id ORDER BY gp_segment_id;" ``` 次のように6つのセグメント（0-5）にデータが分散されていることを確認してください。 ``` gp_segment_id | count ---------------+------- 0 | 172 1 | 163 2 | 182 3 | 164 4 | 168 5 | 151 (6 rows) ``` --- 以上でセットアップ完了です。