refactor(mis,portal): 重构scow后端, 对接调度器适配器接口 (#632)

### 1. 部署调度器适配器首先需要确保您的集群上部署了对应的调度器适配器，得到访问它的地址及端口号部署适配器可参考文档： - [slurm](https://github.com/PKUHPC/scow-slurm-adapter) ### 2. 修改SCOW配置文件首先确保您使用了最新的SCOW镜像（可查看`install.yaml`中的`imageTag`字段）在用于部署scow的`scow-deployment`文件夹中，修改配置文件： - 首先修改集群配置文件主要变化为删除`slurm`配置项, 将`loginNodes`配置项作为独立的一项配置。新增`adapterUrl`配置项，标识适配器地址 ``` # 集群显示名称 displayName: hpc01 # 调度器适配器的地址 adapterUrl: "192.168.88.101:8972" # 登录节点 loginNodes: - "192.168.88.102" ``` - 修改管理系统配置文件删除了`fetchJobs`配置项中的`db`项，即不再采用源作业信息数据库，通过适配器同步作业信息 ### 3. 不再使用源作业信息数据库部署使用适配器后，可以不再部署[`export-jobs`](https://github.com/PKUHPC/export-jobs)项目，同步作业信息的功能由适配器完成
PKUHPC · Jul 11, 2023 · 5b7f0e8 · 5b7f0e8
1 parent 6b6f08a
commit 5b7f0e8
Show file tree

Hide file tree

Showing 136 changed files with 4,202 additions and 3,941 deletions.
diff --git a/.changeset/afraid-maps-deliver.md b/.changeset/afraid-maps-deliver.md
@@ -0,0 +1,19 @@
+---
+"@scow/scheduler-adapter-protos": minor
+"@scow/lib-scheduler-adapter": minor
+"@scow/portal-server": minor
+"@scow/test-adapter": minor
+"@scow/protos": minor
+"@scow/mis-server": minor
+"@scow/portal-web": minor
+"@scow/demo-vagrant": minor
+"@scow/mis-web": minor
+"@scow/config": minor
+"@scow/auth": minor
+"@scow/cli": minor
+"@scow/lib-ssh": minor
+"@scow/grpc-api": minor
+"@scow/docs": minor
+---
+
+重构 scow，对接调度器适配器接口
diff --git a/.devcontainer/docker-compose.devcontainer.yml b/.devcontainer/docker-compose.devcontainer.yml
@@ -75,6 +75,13 @@ services:
       PHPLDAPADMIN_LDAP_HOSTS: ldap://scow-dev:389
       PHPLDAPADMIN_HTTPS: false
 
+  test-adapter:
+    image: test-adapter
+    build:
+      context: ..
+      dockerfile: dev/test-adapter/Dockerfile
+    network_mode: service:scow-dev
+
 volumes:
   db:
   ldap:

diff --git a/apps/auth/src/auth/ssh/index.ts b/apps/auth/src/auth/ssh/index.ts
@@ -31,7 +31,7 @@ function checkLoginNode(sshConfig: SshConfigSchema) {
     }
     const clusterConfig = Object.values(clusters)[0];
 
-    loginNode = getLoginNode(clusterConfig.slurm.loginNodes[0]).address;
+    loginNode = getLoginNode(clusterConfig.loginNodes[0]).address;
 
     if (!loginNode) {
       throw new Error(`Cluster ${clusterConfig.displayName} has no login node.`);

diff --git a/apps/cli/assets/config/clusters/hpc01.yaml b/apps/cli/assets/config/clusters/hpc01.yaml
@@ -2,79 +2,17 @@
 displayName: hpc01Name
 
 # 指定slurm配置
-slurm:
-  loginNodes:
-    # 登录节点展示名称
-    - name: login01
-      # 登录节点的IP或者域名
-      # 如果设置的是域名，请确认此节点的/etc/hosts中包含了域名到IP的解析信息
-      address: login01
-    - name: login02
-      address: login02
-
-  # 集群的分区信息
-  partitions:
-    # 分区1的名字
-    - name: compute
-      # 分区内节点数
-      nodes: 28
-      # 单节点内存数量，单位M
-      mem: 7500
-      # 核心数
-      cores: 2
-      # GPU卡数
-      gpus: 0
-      # QOS
-      qos:
-        - low
-        - normal
-        - high
-      # 这个分区的备注信息
-      comment: ""
-
-    - name: GPU
-      nodes: 1
-      mem: 262144
-      cores: 48
-      gpus: 8
-      qos:
-        - low
-        - normal
-        - high
-      comment: ""
-
-  # 如果这个集群要部署管理系统，请增加以下配置
-  # 如果不需要，将整个mis块注释掉
-  mis:
-    # 部署slurm.sh的机器的地址
-    managerUrl: haha
-    # slurm.sh在机器中的绝对地址
-    scriptPath: /test/slurm.sh
-
-    # 部署slurm.sh的机器通过什么地址访问slurm的数据库
-    # 不填写为下面的默认值
-    # dbHost: localhost
-
-    # 部署slurm.sh的机器通过什么端口访问slurm的数据库
-    # 不填写为下面的默认值
-    # dbPort: 3306
-
-    # slurm数据库的用户名
-    # 参考slurmdbd.conf的StorageUser配置
-    # 不填写为下面的默认值
-    # dbUser: root
-
-    # slurmdbd的数据库用户的密码
-    # 参考slurmdbd.conf的StoragePass配置
-    dbPassword: password
-
-    # slurm accounting数据库的数据库名
-    # 参考slurmdbd.conf的StorageLoc配置
-    # 不填写为下面的默认值
-    # slurmAcctDbName: "slurm_acct_db"
-
-    # 这个集群在slurm中的集群名字
-    clusterName: pkuhpc
+loginNodes:
+  # 登录节点展示名称
+  - name: login01
+    # 登录节点的IP或者域名
+    # 如果设置的是域名，请确认此节点的/etc/hosts中包含了域名到IP的解析信息
+    address: login01
+  - name: login02
+    address: login02
+
+# 适配器地址(ip地址:端口号)
+adapterUrl: localhost:8972
 
 # 门户系统代理网关节点配置
 # proxyGateway:

diff --git a/apps/cli/assets/config/mis.yaml b/apps/cli/assets/config/mis.yaml
@@ -8,18 +8,6 @@ db:
 
 # 获取作业相关配置
 fetchJobs:
-  # 源作业信息数据库的数据库信息
-  db:
-    host: sourcedb
-    port: 3307
-    user: root
-    password: jobtablepassword
-    dbName: jobs
-    tableName: jobs
-    # 数据库类型，可选mariadb或者mysql
-    # 默认为mariadb
-    # type: mariadb
-
   # 周期性获取数据
   periodicFetch:
     # 是否开启

diff --git a/apps/mis-server/config/clusters/hpc00.yml b/apps/mis-server/config/clusters/hpc00.yml
@@ -1,39 +1,32 @@
 displayName: hpc00
-misIgnore: true
-slurm:
-  mis:
-    managerUrl: localhost:22222
-    dbPassword: password
-    clusterName: pkuhpc
-    scriptPath: /slurmshTest/slurm.sh
+adapterUrl: 0.0.0.0:6000
+loginNodes:
+  - name: login
+    address: localhost:22222
 
-  loginNodes:
-    - name: login
-      address: localhost:22222
-
-  partitions:
-    - name: C032M0128G
-      mem: 131072
-      cores: 32
-      nodes: 32
-      gpus: 0
-      qos:
-        - low
-        - normal
-        - high
-        - cryoem
-    - name: GPU
-      mem: 262144
-      cores: 28
-      nodes: 32
-      gpus: 4
-      qos:
-        - low
-        - normal
-        - high
-        - cryoem
-    - name: life
-      mem: 262144
-      cores: 28
-      gpus: 4
-      nodes: 32
+#   partitions:
+#     - name: C032M0128G
+#       mem: 131072
+#       cores: 32
+#       nodes: 32
+#       gpus: 0
+#       qos:
+#         - low
+#         - normal
+#         - high
+#         - cryoem
+#     - name: GPU
+#       mem: 262144
+#       cores: 28
+#       nodes: 32
+#       gpus: 4
+#       qos:
+#         - low
+#         - normal
+#         - high
+#         - cryoem
+#     - name: life
+#       mem: 262144
+#       cores: 28
+#       gpus: 4
+#       nodes: 32
diff --git a/apps/mis-server/config/clusters/hpc01.yml b/apps/mis-server/config/clusters/hpc01.yml
@@ -1,30 +1,33 @@
 displayName: hpc01
-misIgnore: true
-slurm:
-  mis:
-    managerUrl: localhost:22222
-    dbPassword: password
-    clusterName: pkuhpc1
-    scriptPath: /slurmshTest/slurm.sh
-  loginNodes:
-    - name: login
-      address: localhost:22222
-  partitions:
-    - name: compute
-      nodes: 198
-      mem: 63000
-      cores: 28
-      gpus: 0
-      qos:
-        - low
-        - normal
-        - high
-    - name: gpu
-      nodes: 1
-      mem: 386000
-      cores: 48
-      gpus: 8
-      qos:
-        - low
-        - normal
-        - high
+adapterUrl: 0.0.0.0:6000
+loginNodes:
+  - name: login
+    address: localhost:22222
+# misIgnore: true
+# slurm:
+#   mis:
+#     managerUrl: localhost:22222
+#     dbPassword: password
+#     clusterName: pkuhpc1
+#     scriptPath: /slurmshTest/slurm.sh
+#   loginNodes:
+#     - localhost:22222
+#   partitions:
+#     - name: compute
+#       nodes: 198
+#       mem: 63000
+#       cores: 28
+#       gpus: 0
+#       qos:
+#         - low
+#         - normal
+#         - high
+#     - name: gpu
+#       nodes: 1
+#       mem: 386000
+#       cores: 48
+#       gpus: 8
+#       qos:
+#         - low
+#         - normal
+#         - high
diff --git a/apps/mis-server/config/clusters/hpc02.yml b/apps/mis-server/config/clusters/hpc02.yml
@@ -1,24 +1,27 @@
 displayName: hpc01
-slurm:
-  loginNodes:
-    - name: login
-      address: localhost:22222
-  partitions:
-    - name: compute
-      nodes: 198
-      mem: 63000
-      cores: 28
-      gpus: 0
-      qos:
-        - low
-        - normal
-        - high
-    - name: gpu
-      nodes: 1
-      mem: 386000
-      cores: 48
-      gpus: 8
-      qos:
-        - low
-        - normal
-        - high
+adapterUrl: 0.0.0.0:6000
+loginNodes:
+  - name: login
+    address: localhost:22222
+# slurm:
+#   loginNodes:
+#     - localhost:22222
+#   partitions:
+#     - name: compute
+#       nodes: 198
+#       mem: 63000
+#       cores: 28
+#       gpus: 0
+#       qos:
+#         - low
+#         - normal
+#         - high
+#     - name: gpu
+#       nodes: 1
+#       mem: 386000
+#       cores: 48
+#       gpus: 8
+#       qos:
+#         - low
+#         - normal
+#         - high
diff --git a/apps/mis-server/config/mis.yaml b/apps/mis-server/config/mis.yaml
@@ -5,12 +5,4 @@ db:
   password: mysqlrootpassword
   dbName: scow_server_${JEST_WORKER_ID}
 
-fetchJobs:
-  db:
-    host: 127.0.0.1
-    port: 3307
-    user: root
-    password: jobtablepassword
-    dbName: jobs
-    tableName: jobs
 
diff --git a/apps/mis-server/package.json b/apps/mis-server/package.json
@@ -24,6 +24,7 @@
   "dependencies": {
     "@ddadaal/tsgrpc-common": "0.2.3",
     "@ddadaal/tsgrpc-server": "0.19.2",
+    "@ddadaal/tsgrpc-client": "0.17.5",
     "@grpc/grpc-js": "1.8.15",
     "@mikro-orm/cli": "5.7.12",
     "@mikro-orm/core": "5.7.12",
@@ -40,6 +41,8 @@
     "@scow/lib-slurm": "workspace:*",
     "@scow/lib-ssh": "workspace:*",
     "@scow/protos": "workspace:*",
+    "@scow/scheduler-adapter-protos": "workspace:*",
+    "@scow/lib-scheduler-adapter": "workspace:*",
     "@scow/utils": "workspace:*",
     "@sinclair/typebox": "0.28.15",
     "dotenv": "16.3.1",
@@ -50,7 +53,6 @@
     "wait-on": "7.0.1"
   },
   "devDependencies": {
-    "@ddadaal/tsgrpc-client": "0.17.5",
     "@types/google-protobuf": "3.15.6",
     "@types/node-cron": "3.0.7",
     "@types/wait-on": "5.3.1"

diff --git a/apps/mis-server/src/app.ts b/apps/mis-server/src/app.ts
@@ -19,6 +19,7 @@ import { plugins } from "src/plugins";
 import { accountServiceServer } from "src/services/account";
 import { adminServiceServer } from "src/services/admin";
 import { chargingServiceServer } from "src/services/charging";
+import { configServiceServer } from "src/services/config";
 import { initServiceServer } from "src/services/init";
 import { jobServiceServer } from "src/services/job";
 import { jobChargeLimitServer } from "src/services/jobChargeLimit";
@@ -49,6 +50,7 @@ export async function createServer() {
   await server.register(jobServiceServer);
   await server.register(chargingServiceServer);
   await server.register(tenantServiceServer);
+  await server.register(configServiceServer);
 
   const em = server.ext.orm.em.fork();
   await updateBlockStatusInSlurm(em, server.ext.clusters, server.logger);