generate_ts_config.py 7.81 KB
Newer Older
liuzhe-lz's avatar
liuzhe-lz committed
1
2
# Copyright (c) Microsoft Corporation.
# Licensed under the MIT license.
3

4
import sys
chicm-ms's avatar
chicm-ms committed
5
import os
6
import glob
7
8
9
import argparse
from utils import get_yml_content, dump_yml_content

chicm-ms's avatar
chicm-ms committed
10
TRAINING_SERVICE_FILE = os.path.join('config', 'training_service.yml')
11
12
13
14
15

def update_training_service_config(args):
    config = get_yml_content(TRAINING_SERVICE_FILE)
    if args.nni_manager_ip is not None:
        config[args.ts]['nniManagerIp'] = args.nni_manager_ip
SparkSnail's avatar
SparkSnail committed
16
    if args.ts == 'paiYarn':
17
        if args.pai_user is not None:
18
            config[args.ts]['paiYarnConfig']['userName'] = args.pai_user
19
        if args.pai_pwd is not None:
20
            config[args.ts]['paiYarnConfig']['passWord'] = args.pai_pwd
21
        if args.pai_host is not None:
22
            config[args.ts]['paiYarnConfig']['host'] = args.pai_host
23
24
25
26
27
28
        if args.nni_docker_image is not None:
            config[args.ts]['trial']['image'] = args.nni_docker_image
        if args.data_dir is not None:
            config[args.ts]['trial']['dataDir'] = args.data_dir
        if args.output_dir is not None:
            config[args.ts]['trial']['outputDir'] = args.output_dir
29
30
        if args.vc is not None:
            config[args.ts]['trial']['virtualCluster'] = args.vc
SparkSnail's avatar
SparkSnail committed
31
32
33
34
35
36
37
    if args.ts == 'pai':
        if args.pai_user is not None:
            config[args.ts]['paiConfig']['userName'] = args.pai_user
        if args.pai_host is not None:
            config[args.ts]['paiConfig']['host'] = args.pai_host
        if args.pai_token is not None:
            config[args.ts]['paiConfig']['token'] = args.pai_token
38
39
        if args.pai_reuse is not None:
            config[args.ts]['paiConfig']['reuse'] = args.pai_reuse.lower() == 'true'
SparkSnail's avatar
SparkSnail committed
40
41
        if args.nni_docker_image is not None:
            config[args.ts]['trial']['image'] = args.nni_docker_image
chicm-ms's avatar
chicm-ms committed
42
        if args.nni_manager_nfs_mount_path is not None:
SparkSnail's avatar
SparkSnail committed
43
            config[args.ts]['trial']['nniManagerNFSMountPath'] = args.nni_manager_nfs_mount_path
chicm-ms's avatar
chicm-ms committed
44
        if args.container_nfs_mount_path is not None:
SparkSnail's avatar
SparkSnail committed
45
            config[args.ts]['trial']['containerNFSMountPath'] = args.container_nfs_mount_path
46
47
        if args.pai_storage_config_name is not None:
            config[args.ts]['trial']['paiStorageConfigName'] = args.pai_storage_config_name
SparkSnail's avatar
SparkSnail committed
48
49
        if args.vc is not None:
            config[args.ts]['trial']['virtualCluster'] = args.vc
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
    elif args.ts == 'kubeflow':
        if args.nfs_server is not None:
            config[args.ts]['kubeflowConfig']['nfs']['server'] = args.nfs_server
        if args.nfs_path is not None:
            config[args.ts]['kubeflowConfig']['nfs']['path'] = args.nfs_path
        if args.keyvault_vaultname is not None:
            config[args.ts]['kubeflowConfig']['keyVault']['vaultName'] = args.keyvault_vaultname
        if args.keyvault_name is not None:
            config[args.ts]['kubeflowConfig']['keyVault']['name'] = args.keyvault_name
        if args.azs_account is not None:
            config[args.ts]['kubeflowConfig']['azureStorage']['accountName'] = args.azs_account
        if args.azs_share is not None:
            config[args.ts]['kubeflowConfig']['azureStorage']['azureShare'] = args.azs_share
        if args.nni_docker_image is not None:
            config[args.ts]['trial']['worker']['image'] = args.nni_docker_image
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
    elif args.ts == 'frameworkcontroller':
        if args.nfs_server is not None:
            config[args.ts]['frameworkcontrollerConfig']['nfs']['server'] = args.nfs_server
        if args.nfs_path is not None:
            config[args.ts]['frameworkcontrollerConfig']['nfs']['path'] = args.nfs_path
        if args.keyvault_vaultname is not None:
            config[args.ts]['frameworkcontrollerConfig']['keyVault']['vaultName'] = args.keyvault_vaultname
        if args.keyvault_name is not None:
            config[args.ts]['frameworkcontrollerConfig']['keyVault']['name'] = args.keyvault_name
        if args.azs_account is not None:
            config[args.ts]['frameworkcontrollerConfig']['azureStorage']['accountName'] = args.azs_account
        if args.azs_share is not None:
            config[args.ts]['frameworkcontrollerConfig']['azureStorage']['azureShare'] = args.azs_share
        if args.nni_docker_image is not None:
            config[args.ts]['trial']['taskRoles'][0]['image'] = args.nni_docker_image
SparkSnail's avatar
SparkSnail committed
80
81
82
83
84
85
86
87
88
    elif args.ts == 'remote':
        if args.remote_user is not None:
            config[args.ts]['machineList'][0]['username'] = args.remote_user
        if args.remote_host is not None:
            config[args.ts]['machineList'][0]['ip'] = args.remote_host
        if args.remote_port is not None:
            config[args.ts]['machineList'][0]['port'] = args.remote_port
        if args.remote_pwd is not None:
            config[args.ts]['machineList'][0]['passwd'] = args.remote_pwd
SparkSnail's avatar
SparkSnail committed
89
90
        if args.remote_reuse is not None:
            config[args.ts]['remoteConfig']['reuse'] = args.remote_reuse.lower() == 'true'
SparkSnail's avatar
SparkSnail committed
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
    elif args.ts == 'adl':
        if args.nni_docker_image is not None:
            config[args.ts]['trial']['image'] = args.nni_docker_image
        if args.checkpoint_storage_class is not None:
            config[args.ts]['trial']['checkpoint']['storageClass'] = args.checkpoint_storage_class
        if args.checkpoint_storage_size is not None:
            config[args.ts]['trial']['checkpoint']['storageSize'] = args.checkpoint_storage_size
        if args.adaptive is not None:
            config[args.ts]['trial']['adaptive'] = args.adaptive
        if args.adl_nfs_server is not None and args.adl_nfs_path is not None and args.adl_nfs_container_mount_path is not None:
            # default keys in nfs is empty, need to initialize
            config[args.ts]['trial']['nfs'] = {}
            config[args.ts]['trial']['nfs']['server'] = args.adl_nfs_server
            config[args.ts]['trial']['nfs']['path'] = args.adl_nfs_path
            config[args.ts]['trial']['nfs']['container_mount_path'] = args.nadl_fs_container_mount_path
106
107
108

    dump_yml_content(TRAINING_SERVICE_FILE, config)

109

110
111
if __name__ == '__main__':
    parser = argparse.ArgumentParser()
SparkSnail's avatar
SparkSnail committed
112
    parser.add_argument("--ts", type=str, choices=['pai', 'kubeflow', 'remote', 'local', 'frameworkcontroller', 'adl'], default='pai')
113
114
115
116
117
118
119
120
    parser.add_argument("--nni_docker_image", type=str)
    parser.add_argument("--nni_manager_ip", type=str)
    # args for PAI
    parser.add_argument("--pai_user", type=str)
    parser.add_argument("--pai_pwd", type=str)
    parser.add_argument("--pai_host", type=str)
    parser.add_argument("--data_dir", type=str)
    parser.add_argument("--output_dir", type=str)
121
    parser.add_argument("--vc", type=str)
SparkSnail's avatar
SparkSnail committed
122
    parser.add_argument("--pai_token", type=str)
123
    parser.add_argument("--pai_reuse", type=str)
124
    parser.add_argument("--pai_storage_config_name", type=str)
SparkSnail's avatar
SparkSnail committed
125
126
    parser.add_argument("--nni_manager_nfs_mount_path", type=str)
    parser.add_argument("--container_nfs_mount_path", type=str)
127
    # args for kubeflow and frameworkController
128
129
130
131
132
133
    parser.add_argument("--nfs_server", type=str)
    parser.add_argument("--nfs_path", type=str)
    parser.add_argument("--keyvault_vaultname", type=str)
    parser.add_argument("--keyvault_name", type=str)
    parser.add_argument("--azs_account", type=str)
    parser.add_argument("--azs_share", type=str)
SparkSnail's avatar
SparkSnail committed
134
135
136
137
138
    # args for remote
    parser.add_argument("--remote_user", type=str)
    parser.add_argument("--remote_pwd", type=str)
    parser.add_argument("--remote_host", type=str)
    parser.add_argument("--remote_port", type=int)
SparkSnail's avatar
SparkSnail committed
139
    parser.add_argument("--remote_reuse", type=str)
SparkSnail's avatar
SparkSnail committed
140
141
142
143
144
145
146
    # args for adl
    parser.add_argument("--checkpoint_storage_class", type=str)
    parser.add_argument("--checkpoint_storage_size", type=str)
    parser.add_argument("--adaptive", type=str)
    parser.add_argument("--adl_nfs_server", type=str)
    parser.add_argument("--adl_nfs_path", type=str)
    parser.add_argument("--adl_nfs_container_mount_path", type=str)
147
148
149
    args = parser.parse_args()

    update_training_service_config(args)