Create sync_data.sh
Browse files- sync_data.sh +107 -0
sync_data.sh
ADDED
@@ -0,0 +1,107 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#!/bin/sh
|
2 |
+
|
3 |
+
# 检查环境变量
|
4 |
+
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
|
5 |
+
echo "Starting without backup functionality - missing HF_TOKEN or DATASET_ID"
|
6 |
+
exec node ./src/app/app.js
|
7 |
+
exit 0
|
8 |
+
fi
|
9 |
+
|
10 |
+
# 激活虚拟环境
|
11 |
+
. /opt/venv/bin/activate
|
12 |
+
|
13 |
+
# 上传备份
|
14 |
+
upload_backup() {
|
15 |
+
file_path="$1"
|
16 |
+
file_name="$2"
|
17 |
+
|
18 |
+
python3 -c "
|
19 |
+
from huggingface_hub import HfApi
|
20 |
+
import sys
|
21 |
+
import os
|
22 |
+
api = HfApi(token='$HF_TOKEN')
|
23 |
+
try:
|
24 |
+
api.upload_file(
|
25 |
+
path_or_fileobj='$file_path',
|
26 |
+
path_in_repo='$file_name',
|
27 |
+
repo_id='$DATASET_ID',
|
28 |
+
repo_type='dataset'
|
29 |
+
)
|
30 |
+
print(f'Successfully uploaded $file_name')
|
31 |
+
except Exception as e:
|
32 |
+
print(f'Error uploading file: {str(e)}')
|
33 |
+
"
|
34 |
+
}
|
35 |
+
|
36 |
+
# 下载最新备份
|
37 |
+
download_latest_backup() {
|
38 |
+
python3 -c "
|
39 |
+
from huggingface_hub import HfApi
|
40 |
+
import sys
|
41 |
+
import os
|
42 |
+
import tarfile
|
43 |
+
import tempfile
|
44 |
+
api = HfApi(token='$HF_TOKEN')
|
45 |
+
try:
|
46 |
+
files = api.list_repo_files(repo_id='$DATASET_ID', repo_type='dataset')
|
47 |
+
backup_files = [f for f in files if f.startswith('electerm_backup_') and f.endswith('.tar.gz')]
|
48 |
+
|
49 |
+
if not backup_files:
|
50 |
+
print('No backup files found')
|
51 |
+
sys.exit()
|
52 |
+
|
53 |
+
latest_backup = sorted(backup_files)[-1]
|
54 |
+
|
55 |
+
with tempfile.TemporaryDirectory() as temp_dir:
|
56 |
+
filepath = api.hf_hub_download(
|
57 |
+
repo_id='$DATASET_ID',
|
58 |
+
filename=latest_backup,
|
59 |
+
repo_type='dataset',
|
60 |
+
local_dir=temp_dir
|
61 |
+
)
|
62 |
+
|
63 |
+
if filepath and os.path.exists(filepath):
|
64 |
+
with tarfile.open(filepath, 'r:gz') as tar:
|
65 |
+
tar.extractall('/app/electerm-web/data')
|
66 |
+
print(f'Successfully restored backup from {latest_backup}')
|
67 |
+
|
68 |
+
except Exception as e:
|
69 |
+
print(f'Error downloading backup: {str(e)}')
|
70 |
+
"
|
71 |
+
}
|
72 |
+
|
73 |
+
# 首次启动时下载最新备份
|
74 |
+
echo "Downloading latest backup from HuggingFace..."
|
75 |
+
download_latest_backup
|
76 |
+
|
77 |
+
# 同步函数
|
78 |
+
sync_data() {
|
79 |
+
while true; do
|
80 |
+
echo "Starting sync process at $(date)"
|
81 |
+
|
82 |
+
if [ -d /app/electerm-web/data ]; then
|
83 |
+
timestamp=$(date +%Y%m%d_%H%M%S)
|
84 |
+
backup_file="electerm_backup_${timestamp}.tar.gz"
|
85 |
+
|
86 |
+
# 压缩数据目录
|
87 |
+
tar -czf "/tmp/${backup_file}" -C /app/electerm-web/data .
|
88 |
+
|
89 |
+
echo "Uploading backup to HuggingFace..."
|
90 |
+
upload_backup "/tmp/${backup_file}" "${backup_file}"
|
91 |
+
|
92 |
+
rm -f "/tmp/${backup_file}"
|
93 |
+
else
|
94 |
+
echo "Data directory does not exist yet, waiting for next sync..."
|
95 |
+
fi
|
96 |
+
|
97 |
+
SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
|
98 |
+
echo "Next sync in ${SYNC_INTERVAL} seconds..."
|
99 |
+
sleep $SYNC_INTERVAL
|
100 |
+
done
|
101 |
+
}
|
102 |
+
|
103 |
+
# 后台启动同步进程
|
104 |
+
sync_data &
|
105 |
+
|
106 |
+
# 启动 Electerm
|
107 |
+
exec node ./src/app/app.js
|