StarRocks 運維工具 StarGo
- 2022 年 8 月 25 日
- 筆記
註:本文主要內容均來源 StarRocks 官網 //docs.starrocks.com/zh-cn/main/administration/stargo
StarGo 是一個用於管理多個 StarRocks 集群的命令行工具。通過 StarGo,您可以使用簡單的命令行實現多集群的部署、查看、升級、啟動與停止等操作。
StarGo 使用 go 開發,github 地址 //github.com/wangtianyi2004/starrocks-controller
部署 StarGo
在當前用戶路徑下下載並解壓 StarGo 二進制安裝包。
wget //raw.githubusercontent.com/wangtianyi2004/starrocks-controller/main/stargo-pkg.tar.gz
tar -xzvf stargo-pkg.tar.gz
安裝包包含以下文件。
stargo:StarGo 二進制文件,無需安裝。
deploy-template.yaml:部署配置文件模板。
repo.yaml:指定 StarRocks 安裝包下載庫的配置文件。
- 說明:您可以在 //cdn-thirdparty.starrocks.com 地址下獲取相應的版本的安裝 index 文件以及安裝包。
StarGo 支持的版本參見: //starrocks-thirdparty.oss-cn-zhangjiakou.aliyuncs.com/packageVersion.list
目前支持版本如下:
v2.0.1
v2.1.3
v2.1.6
v2.2.0
v2.2.1
v2.2.2
v2.2.3
v2.3.0
部署集群
前提條件
- 待部署集群至少需要一個中控機節點和三個部署機節點,所有節點可以混合部署於同一台機器。
- 中控機上需部署 StarGo。
- 中控機與部署機間需創建 SSH 互信。
以下示例創建了中控機 sr-dev@r0 與部署機 starrocks@r1、starrocks@r2 以及 starrocks@r3 間的 SSH 互信。
創建 sr-dev@r0 到 starrocks@r1、r2、r3 的 ssh 互信。
[sr-dev@r0 ~]$ ssh-keygen
[sr-dev@r0 ~]$ ssh-copy-id starrocks@r1
[sr-dev@r0 ~]$ ssh-copy-id starrocks@r2
[sr-dev@r0 ~]$ ssh-copy-id starrocks@r3
## 驗證 sr-dev@r0 到 starrocks@r1、r2、r3 的 ssh 互信。
[sr-dev@r0 ~]$ ssh starrocks@r1 date
[sr-dev@r0 ~]$ ssh starrocks@r2 date
[sr-dev@r0 ~]$ ssh starrocks@r3 date
創建配置文件
根據以下 YAML 模板,創建部署 StarRocks 集群的拓撲文件。具體配置項參考參數配置。
[starrocks@bigdata11 stargo-pkg]$ cat deploy-star3.yaml
global:
user: "starrocks"
ssh_port: 22
fe_servers:
- host: xx.xx.xx.229
ssh_port: 22
http_port: 38030
rpc_port: 39020
query_port: 39030
edit_log_port: 39010
deploy_dir: /opt/StarRocks1/fe
meta_dir: /data/starrocks1/fe/meta
log_dir: /data/starrocks1/fe/log
priority_networks: xx.xx.xx.229
config:
sys_log_level: "INFO"
be_servers:
- host: xx.xx.xx.229
ssh_port: 22
be_port: 39060
webserver_port: 38040
heartbeat_service_port: 39050
brpc_port: 38060
deploy_dir : /opt/StarRocks1/be
storage_dir: /data/starrocks1/be/storage
log_dir: /data/starrocks1/be/log
priority_networks: xx.xx.xx.229/24
config:
create_tablet_worker_count: 3
- 註:服務器上已經有一套 StarRocks,再部署一套單機的,修改了下程序、端口
創建部署目錄(可選)
如果您在配置文件中設定的部署路徑不存在,且您有創建該路徑的權限,StarGo 將根據配置文件自動創建部署目錄。如果路徑已存在,請確保您有在該路徑下擁有寫入的權限。您也可以通過以下命令,在各部署節點分別創建部署路徑。
- 在 FE 節點安裝目錄下上創建 meta 路徑。
mkdir -p /opt/StarRocks1/be
- 在 BE 節點安裝目錄下上創建 storage 路徑。
mkdir -p /data/starrocks1/be/storage
- 注意: 請確保以上創建的路徑與配置文件中的 meta_dir 和 storage_dir 相同。
部署 StarRocks
通過以下命令部署 StarRocks 集群。
./stargo cluster deploy <cluster_name> <version> <topology_file>
| 參數 | 描述 |
|---|---|
| cluster_name | 創建的集群名 |
| version | StarRocks 的版本 |
| topology_file | 配置文件名 |
創建成功後,集群將會自動啟動。當返回 beStatus 和feStatus 為 true 時,集群部署啟動成功。
示例:
[starrocks@bigdata11 stargo-pkg]$ ./stargo cluster deploy star3 v2.2.2 deploy-star3.yaml
[20220812-113235 OUTPUT] Deploy cluster [clusterName = star3, clusterVersion = v2.2.2, metaFile = deploy-star3.yaml]
[20220812-113236 OUTPUT] PRE CHECK DEPLOY ENV:
PreCheck FE:
server id ssh auth meta dir deploy dir http port rpc port query port edit log port open files count
-------------------- --------------- ------------------------------ ------------------------------ --------------- --------------- --------------- --------------- ---------------
xx.xx.xx.229:39010 PASS PASS PASS PASS PASS PASS PASS PASS
PreCheck BE:
server id ssh auth storage dir deploy dir webSer port heartbeat port brpc port be port open files count
-------------------- --------------- ------------------------------ ------------------------------ --------------- --------------- --------------- --------------- ---------------
xx.xx.xx.229:39060 PASS PASS PASS PASS PASS PASS PASS PASS
[20220812-113236 OUTPUT] PreCheck successfully. RESPECT
[20220812-113236 OUTPUT] Create the deploy folder ...
[20220812-113237 OUTPUT] Download StarRocks package & jdk ...
[20220812-113322 INFO] The file starrocks-2.2.2-quickstart.tar.gz [1695364308] download successfully
[20220812-113322 OUTPUT] Download done.
[20220812-113322 OUTPUT] Decompress StarRocks pakcage & jdk ...
[20220812-113325 INFO] The tar file /home/starrocks/.stargo/download/starrocks-2.2.2-quickstart.tar.gz has been decompressed under /home/starrocks/.stargo/download
[20220812-113349 INFO] The tar file /home/starrocks/.stargo/download/StarRocks-2.2.2.tar.gz has been decompressed under /home/starrocks/.stargo/download
[20220812-113354 INFO] The tar file /home/starrocks/.stargo/download/jdk-8u301-linux-x64.tar.gz has been decompressed under /home/starrocks/.stargo/download
[20220812-113354 OUTPUT] Distribute FE Dir ...
[20220812-113402 INFO] Upload dir feSourceDir = [/home/starrocks/.stargo/download/StarRocks-2.2.2/fe] to feTargetDir = [/opt/StarRocks1/fe] on FeHost = [xx.xx.xx.229]
[20220812-113406 INFO] Upload dir JDKSourceDir = [/home/starrocks/.stargo/download/jdk1.8.0_301] to JDKTargetDir = [/opt/StarRocks1/fe/jdk] on FeHost = [xx.xx.xx.229]
[20220812-113406 INFO] Modify JAVA_HOME: host = [xx.xx.xx.229], filePath = [/opt/StarRocks1/fe/bin/start_fe.sh]
[20220812-113406 OUTPUT] Distribute BE Dir ...
[20220812-113417 INFO] Upload dir BeSourceDir = [/home/starrocks/.stargo/download/StarRocks-2.2.2/be] to BeTargetDir = [/opt/StarRocks1/be] on BeHost = [xx.xx.xx.229]
[20220812-113417 OUTPUT] Modify configuration for FE nodes & BE nodes ...
############################################# START FE CLUSTER #############################################
############################################# START FE CLUSTER #############################################
[20220812-113417 INFO] Starting leader FE node [host = xx.xx.xx.229, editLogPort = 39010]
[20220812-113438 WARN] The FE node doesn't start, wait for 10s [FeHost = xx.xx.xx.229, FeQueryPort = 39030, error = Process exited with status 1]
[20220812-113438 INFO] Starting leader FE node [host = xx.xx.xx.229, editLogPort = 39010]
[20220812-113454 INFO] The FE node start succefully [host = xx.xx.xx.229, queryPort = 39030]
[20220812-113454 INFO] List all FE status:
feHost = xx.xx.xx.229 feQueryPort = 39030 feStatus = true
############################################# START BE CLUSTER #############################################
############################################# START BE CLUSTER #############################################
[20220812-113454 INFO] Starting BE node [BeHost = xx.xx.xx.229 HeartbeatServicePort = 39050]
[20220812-113515 INFO] The BE node start succefully [host = xx.xx.xx.229, heartbeatServicePort = 39050]
[20220812-113515 OUTPUT] List all BE status:
beHost = xx.xx.xx.229 beHeartbeatServicePort = 39050 beStatus = true
如果執行檢查失敗,會有相應提示,按提示操作即可
[starrocks@bigdata12 stargo-pkg]$ ./stargo cluster deploy star3 v2.2.2 deploy-star3.yaml
[20220812-102653 OUTPUT] Deploy cluster [clusterName = star3, clusterVersion = v2.2.2, metaFile = deploy-star3.yaml]
[20220812-102658 OUTPUT] PRE CHECK DEPLOY ENV:
PreCheck FE:
server id ssh auth meta dir deploy dir http port rpc port query port edit log port open files count
-------------------- --------------- ------------------------------ ------------------------------ --------------- --------------- --------------- --------------- ---------------
xx.xx.xx.228:39010 PASS FAILED: Priv failed FAILED: Priv failed PASS PASS PASS PASS PASS
PreCheck BE:
server id ssh auth storage dir deploy dir webSer port heartbeat port brpc port be port open files count
-------------------- --------------- ------------------------------ ------------------------------ --------------- --------------- --------------- --------------- ---------------
xx.xx.xx.228:39060 PASS FAILED: Dir exist/Priv failed FAILED: Dir exist/Priv failed PASS PASS PASS PASS PASS
xx.xx.xx.229:39060 PASS FAILED: Dir exist/Priv failed FAILED: Dir exist/Priv failed PASS PASS PASS PASS PASS
xx.xx.xx.230:39060 PASS FAILED: Dir exist FAILED: Dir exist PASS PASS PASS PASS PASS
[20220812-102658 ERROR] Please use bellowing promption to fix the issue for FE servers:
Detect the FE META FOLDER exist or no privilege. Use bellowing command to check or fix the issue:
[Host = xx.xx.xx.228] chown -R starrocks /data/starrocks1/fe/meta
Detect the FE DEPLOY FOLDER exist or no privilege. Use bellowing command to check or fix the issue:
[Host = xx.xx.xx.228] chown -R starrocks /opt/StarRocks1/fe
[20220812-102658 ERROR] Please use bellowing promption to fix the issue for BE servers:
Detect the BE STORAGE FOLDER exist or no privilege. Use bellowing command to check or fix the issue:
[Host = xx.xx.xx.228] mkdir /data/starrocks1/be/storage.bak && mv /data/starrocks1/be/storage/* /data/starrocks1/be/storage.bak/ && chown -R starrocks /data/starrocks1/be/storage
[Host = xx.xx.xx.229] mkdir /data/starrocks1/be/storage.bak && mv /data/starrocks1/be/storage/* /data/starrocks1/be/storage.bak/ && chown -R starrocks /data/starrocks1/be/storage
[Host = xx.xx.xx.230] mkdir /data/starrocks1/be/storage.bak && mv /data/starrocks1/be/storage/* /data/starrocks1/be/storage.bak/
Detect the BE DEPLOY FOLDER exist or no privilege. Use bellowing command to check or fix the issue:
[Host = xx.xx.xx.228] mkdir /opt/StarRocks1/be.bak && mv /opt/StarRocks1/be/* /opt/StarRocks1/be.bak/ && chown -R starrocks /opt/StarRocks1/be
[Host = xx.xx.xx.229] mkdir /opt/StarRocks1/be.bak && mv /opt/StarRocks1/be/* /opt/StarRocks1/be.bak/ && chown -R starrocks /opt/StarRocks1/be
[Host = xx.xx.xx.230] mkdir /opt/StarRocks1/be.bak && mv /opt/StarRocks1/be/* /opt/StarRocks1/be.bak/
[20220812-102658 ERROR] PreCheck failed.
查看集群信息
[starrocks@bigdata11 stargo-pkg]$ ./stargo cluster list
[20220812-115246 OUTPUT] List all clusters
ClusterName Version User CreateDate MetaPath PrivateKey
--------------- ---------- ---------- ------------------------- ------------------------------------------------------------ --------------------------------------------------
star3 v2.2.2 starrocks 2022-08-12 11:34:17 /home/starrocks/.stargo/cluster/star3 /home/starrocks/.ssh/id_rsa
查看指定集群信息
[starrocks@bigdata11 stargo-pkg]$ ./stargo cluster display star3
[20220812-143023 OUTPUT] Display cluster [clusterName = star3]
clusterName = star3
clusterVerison = v2.2.2
ID ROLE HOST PORT STAT DATADIR DEPLOYDIR
-------------------------- ------ -------------------- --------------- ---------- -------------------------------------------------- --------------------------------------------------
xx.xx.xx.229:39010 FE xx.xx.xx.229 39010/39030 UP/L /opt/StarRocks1/fe /data/starrocks1/fe/meta
xx.xx.xx.229:39060 BE xx.xx.xx.229 39060/39050 UP /opt/StarRocks1/be /data/starrocks1/be/storage
[starrocks@bigdata11 stargo-pkg]$
升級版本
starrocks 2.2.0 升級到 2.3.0
[starrocks@bigdata11 stargo-pkg]$ ./stargo cluster upgrade star3 v2.3.0
[20220812-143227 OUTPUT] Upgrade cluster. [ClusterName = star3, TargetVersion = v2.3.0]
[20220812-143227 OUTPUT] Upgrade StarRocks Cluster star3, from version v2.2.2 to version v2.3.0
[20220812-143227 OUTPUT] Download StarRocks package & jdk ...
[20220812-143340 INFO] The file starrocks-2.3.0-quickstart.tar.gz [1726845129] download successfully
[20220812-143340 OUTPUT] Download done.
[20220812-143340 OUTPUT] Decompress StarRocks pakcage & jdk ...
[20220812-143346 INFO] The tar file /home/starrocks/.stargo/download/starrocks-2.3.0-quickstart.tar.gz has been decompressed under /home/starrocks/.stargo/download
[20220812-143411 INFO] The tar file /home/starrocks/.stargo/download/StarRocks-2.3.0.tar.gz has been decompressed under /home/starrocks/.stargo/download
[20220812-143411 INFO] The tar file /home/starrocks/.stargo/download/jdk-8u301-linux-x64.tar.gz has been decompressed under /home/starrocks/.stargo/download
[20220812-143413 OUTPUT] Starting upgrade BE node. [beId = 0]
[20220812-143413 INFO] upgrade be node - backup be lib. [host = xx.xx.xx.229, sourceDir = /opt/StarRocks1/be/lib, targetDir = /opt/StarRocks1/be/lib.bak-20220812143413]
[20220812-143423 INFO] upgrade be node - upload new be lib. [host = xx.xx.xx.229, sourceDir = /home/starrocks/.stargo/download/StarRocks-2.3.0/be/lib, targetDir = /opt/StarRocks1/be/lib]
[20220812-143423 INFO] Waiting for stoping BE node [BeHost = xx.xx.xx.229]
[20220812-143429 INFO] upgrade be node - stop be node. [host = xx.xx.xx.229, beDeployDir = /opt/StarRocks1/be]
[20220812-143430 INFO] upgrade be node - start be node. [host = xx.xx.xx.229, beDeployDir = /opt/StarRocks1/be]
[20220812-143443 INFO] upgrade be node - start be node. [host = xx.xx.xx.229, beDeployDir = /opt/StarRocks1/be]
[20220812-143443 OUTPUT] The Be node upgrade successfully. [beId = 0, currentVersion = v2.3.0-a9bdb09]
[20220812-143443 OUTPUT] Starting upgrade FE node. [feId = 0]
[20220812-143445 INFO] upgrade FE node - backup FE lib. [host = xx.xx.xx.229, sourceDir = /opt/StarRocks1/fe/lib, targetDir = /opt/StarRocks1/fe/lib.bak-20220812143443]
[20220812-143448 INFO] upgrade FE node - upload new FE lib. [host = xx.xx.xx.229, sourceDir = /home/starrocks/.stargo/download/StarRocks-2.3.0/fe/lib, targetDir = /opt/StarRocks1/fe/lib]
[20220812-143448 INFO] Waiting for stoping FE node [FeHost = xx.xx.xx.229]
[20220812-143451 INFO] upgrade FE node - stop FE node. [host = xx.xx.xx.229, feDeployDir = /opt/StarRocks1/fe]
[20220812-143452 INFO] upgrade FE node - start FE node. [host = xx.xx.xx.229, feDeployDir = /opt/StarRocks1/fe]
[20220812-143452 ERROR] Error in ping db [dbPath = root:@tcp(xx.xx.xx.229:39030)/], error = dial tcp xx.xx.xx.229:39030: connect: connection refused
[20220812-143502 INFO] upgrade FE node - start FE node. [host = xx.xx.xx.229, feDeployDir = /opt/StarRocks1/fe]
[20220812-143503 ERROR] The FE node upgrade failed. [feId = 0, targetVersion = v2.3.0, currentVersion = v]
- 註: 是有報錯,但是升級成功了
升級後查看集群
[20220812-143555 OUTPUT] List all clusters
ClusterName Version User CreateDate MetaPath PrivateKey
--------------- ---------- ---------- ------------------------- ------------------------------------------------------------ --------------------------------------------------
star3 v2.3.0 starrocks 2022-08-12 14:35:03 /home/starrocks/.stargo/cluster/star3 /home/starrocks/.ssh/id_rsa
已知缺陷
- StarGo 不會部署 Broker
- StarGo 安裝包包含 jdk,不會替換安裝
歡迎關注Flink菜鳥公眾號,會不定期更新Flink(開發技術)相關的推文



