Update sync_data.sh
Browse files- sync_data.sh +90 -40
sync_data.sh
CHANGED
@@ -1,13 +1,23 @@
|
|
1 |
#!/bin/sh
|
2 |
|
3 |
-
#
|
4 |
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
|
5 |
-
echo "
|
6 |
exit 1
|
7 |
fi
|
8 |
|
9 |
# 激活虚拟环境
|
10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
# 生成同步脚本
|
13 |
cat > hf_sync.py << 'EOL'
|
@@ -46,7 +56,8 @@ def upload_backup(file_path, file_name, token, repo_id):
|
|
46 |
manage_backups(api, repo_id)
|
47 |
except Exception as e:
|
48 |
print(f"上传文件出错: {str(e)}")
|
49 |
-
|
|
|
50 |
|
51 |
# 下载最新备份
|
52 |
def download_latest_backup(token, repo_id, extract_path):
|
@@ -56,7 +67,7 @@ def download_latest_backup(token, repo_id, extract_path):
|
|
56 |
backup_files = [f for f in files if f.startswith('backup_') and f.endswith('.tar.gz')]
|
57 |
if not backup_files:
|
58 |
print("未找到任何备份文件")
|
59 |
-
return
|
60 |
latest_backup = sorted(backup_files)[-1]
|
61 |
with tempfile.TemporaryDirectory() as temp_dir:
|
62 |
filepath = api.hf_hub_download(
|
@@ -69,8 +80,10 @@ def download_latest_backup(token, repo_id, extract_path):
|
|
69 |
with tarfile.open(filepath, 'r:gz') as tar:
|
70 |
tar.extractall(extract_path)
|
71 |
print(f"已成功恢复备份: {latest_backup}")
|
|
|
72 |
except Exception as e:
|
73 |
print(f"下载备份出错: {str(e)}")
|
|
|
74 |
|
75 |
# 合并历史提交
|
76 |
def super_squash_history(token, repo_id):
|
@@ -80,77 +93,114 @@ def super_squash_history(token, repo_id):
|
|
80 |
print("历史合并完成。")
|
81 |
except Exception as e:
|
82 |
print(f"合并历史出错: {str(e)}")
|
|
|
|
|
83 |
|
84 |
# 主函数
|
85 |
if __name__ == "__main__":
|
|
|
|
|
|
|
|
|
86 |
action = sys.argv[1]
|
87 |
token = sys.argv[2]
|
88 |
repo_id = sys.argv[3]
|
|
|
89 |
if action == "upload":
|
|
|
|
|
|
|
90 |
file_path = sys.argv[4]
|
91 |
file_name = sys.argv[5]
|
92 |
-
upload_backup(file_path, file_name, token, repo_id)
|
|
|
93 |
elif action == "download":
|
94 |
extract_path = sys.argv[4] if len(sys.argv) > 4 else '.'
|
95 |
-
download_latest_backup(token, repo_id, extract_path)
|
|
|
96 |
elif action == "super_squash":
|
97 |
-
super_squash_history(token, repo_id)
|
|
|
|
|
|
|
|
|
|
|
|
|
98 |
EOL
|
99 |
-
|
|
|
100 |
echo "正在从 HuggingFace 下载最新备份..."
|
101 |
-
python hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}" "
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
102 |
|
103 |
# 同步函数
|
104 |
sync_data() {
|
105 |
while true; do
|
106 |
echo "同步进程启动于 $(date)"
|
107 |
|
108 |
-
#
|
109 |
-
STORAGE_PATH="/OpenList/data"
|
110 |
if [ -d "${STORAGE_PATH}" ]; then
|
111 |
# 创建备份
|
112 |
timestamp=$(date +%Y%m%d_%H%M%S)
|
113 |
backup_file="backup_${timestamp}.tar.gz"
|
|
|
114 |
|
115 |
-
#
|
116 |
-
|
117 |
-
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
|
122 |
-
|
123 |
-
|
124 |
-
|
125 |
-
|
126 |
-
|
127 |
-
|
128 |
-
|
129 |
-
|
130 |
-
|
131 |
-
|
132 |
-
|
133 |
-
|
134 |
-
|
135 |
-
|
136 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
137 |
else
|
138 |
-
echo "
|
139 |
fi
|
140 |
-
fi
|
141 |
|
142 |
# 清理临时文件
|
143 |
-
rm -f "
|
144 |
else
|
145 |
-
echo "存储目录 ${STORAGE_PATH} 不存在,等待中..."
|
|
|
|
|
|
|
|
|
|
|
|
|
146 |
fi
|
147 |
|
148 |
# 同步间隔
|
149 |
-
SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
|
150 |
echo "下次同步将在 ${SYNC_INTERVAL} 秒后进行..."
|
151 |
sleep $SYNC_INTERVAL
|
152 |
done
|
153 |
}
|
154 |
|
155 |
# 启动同步进程
|
156 |
-
sync_data
|
|
|
1 |
#!/bin/sh
|
2 |
|
3 |
+
# 检查必要的环境变量
|
4 |
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
|
5 |
+
echo "错误: 未设置 HF_TOKEN 或 DATASET_ID 环境变量" >&2
|
6 |
exit 1
|
7 |
fi
|
8 |
|
9 |
# 激活虚拟环境
|
10 |
+
if [ -f "$APP_HOME/venv/bin/activate" ]; then
|
11 |
+
. "$APP_HOME/venv/bin/activate"
|
12 |
+
else
|
13 |
+
echo "错误: 虚拟环境激活脚本未找到" >&2
|
14 |
+
exit 1
|
15 |
+
fi
|
16 |
+
|
17 |
+
# 设置默认同步间隔
|
18 |
+
SYNC_INTERVAL=${SYNC_INTERVAL:-7200} # 默认2小时
|
19 |
+
STORAGE_PATH="$APP_HOME/data"
|
20 |
+
SQUASH_FLAG_FILE="/tmp/last_squash_time"
|
21 |
|
22 |
# 生成同步脚本
|
23 |
cat > hf_sync.py << 'EOL'
|
|
|
56 |
manage_backups(api, repo_id)
|
57 |
except Exception as e:
|
58 |
print(f"上传文件出错: {str(e)}")
|
59 |
+
return False
|
60 |
+
return True
|
61 |
|
62 |
# 下载最新备份
|
63 |
def download_latest_backup(token, repo_id, extract_path):
|
|
|
67 |
backup_files = [f for f in files if f.startswith('backup_') and f.endswith('.tar.gz')]
|
68 |
if not backup_files:
|
69 |
print("未找到任何备份文件")
|
70 |
+
return False
|
71 |
latest_backup = sorted(backup_files)[-1]
|
72 |
with tempfile.TemporaryDirectory() as temp_dir:
|
73 |
filepath = api.hf_hub_download(
|
|
|
80 |
with tarfile.open(filepath, 'r:gz') as tar:
|
81 |
tar.extractall(extract_path)
|
82 |
print(f"已成功恢复备份: {latest_backup}")
|
83 |
+
return True
|
84 |
except Exception as e:
|
85 |
print(f"下载备份出错: {str(e)}")
|
86 |
+
return False
|
87 |
|
88 |
# 合并历史提交
|
89 |
def super_squash_history(token, repo_id):
|
|
|
93 |
print("历史合并完成。")
|
94 |
except Exception as e:
|
95 |
print(f"合并历史出错: {str(e)}")
|
96 |
+
return False
|
97 |
+
return True
|
98 |
|
99 |
# 主函数
|
100 |
if __name__ == "__main__":
|
101 |
+
if len(sys.argv) < 4:
|
102 |
+
print("错误: 参数不足")
|
103 |
+
sys.exit(1)
|
104 |
+
|
105 |
action = sys.argv[1]
|
106 |
token = sys.argv[2]
|
107 |
repo_id = sys.argv[3]
|
108 |
+
|
109 |
if action == "upload":
|
110 |
+
if len(sys.argv) < 6:
|
111 |
+
print("错误: upload 操作需要文件路径和文件名参数")
|
112 |
+
sys.exit(1)
|
113 |
file_path = sys.argv[4]
|
114 |
file_name = sys.argv[5]
|
115 |
+
if not upload_backup(file_path, file_name, token, repo_id):
|
116 |
+
sys.exit(1)
|
117 |
elif action == "download":
|
118 |
extract_path = sys.argv[4] if len(sys.argv) > 4 else '.'
|
119 |
+
if not download_latest_backup(token, repo_id, extract_path):
|
120 |
+
sys.exit(1)
|
121 |
elif action == "super_squash":
|
122 |
+
if not super_squash_history(token, repo_id):
|
123 |
+
sys.exit(1)
|
124 |
+
else:
|
125 |
+
print(f"错误: 未知操作 '{action}'")
|
126 |
+
sys.exit(1)
|
127 |
+
|
128 |
+
sys.exit(0)
|
129 |
EOL
|
130 |
+
|
131 |
+
# 首次启动时从 HuggingFace 下载最新备份
|
132 |
echo "正在从 HuggingFace 下载最新备份..."
|
133 |
+
if ! python hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}" "${APP_HOME}"; then
|
134 |
+
echo "警告: 下载备份失败,将使用空数据目录启动" >&2
|
135 |
+
fi
|
136 |
+
|
137 |
+
# 启动 OpenList 服务
|
138 |
+
echo "启动 OpenList 服务..."
|
139 |
+
$APP_HOME/openlist server &
|
140 |
+
OPENLIST_PID=$!
|
141 |
|
142 |
# 同步函数
|
143 |
sync_data() {
|
144 |
while true; do
|
145 |
echo "同步进程启动于 $(date)"
|
146 |
|
147 |
+
# 确保数据目录存在
|
|
|
148 |
if [ -d "${STORAGE_PATH}" ]; then
|
149 |
# 创建备份
|
150 |
timestamp=$(date +%Y%m%d_%H%M%S)
|
151 |
backup_file="backup_${timestamp}.tar.gz"
|
152 |
+
temp_backup="/tmp/${backup_file}"
|
153 |
|
154 |
+
# 压缩目录
|
155 |
+
echo "正在创建数据备份..."
|
156 |
+
if tar -czf "${temp_backup}" -C "$(dirname "${STORAGE_PATH}")" "$(basename "${STORAGE_PATH}")"; then
|
157 |
+
# 上传到 HuggingFace
|
158 |
+
echo "正在上传备份到 HuggingFace..."
|
159 |
+
if python hf_sync.py upload "${HF_TOKEN}" "${DATASET_ID}" "${temp_backup}" "${backup_file}"; then
|
160 |
+
echo "备份上传成功"
|
161 |
+
else
|
162 |
+
echo "错误: 备份上传失败" >&2
|
163 |
+
fi
|
164 |
+
|
165 |
+
# 合并历史提交
|
166 |
+
NOW=$(date +%s)
|
167 |
+
SEVEN_DAYS=$((7*24*60*60))
|
168 |
+
if [ ! -f "$SQUASH_FLAG_FILE" ]; then
|
169 |
+
echo $NOW > "$SQUASH_FLAG_FILE"
|
170 |
+
echo "首次合并历史提交..."
|
171 |
+
python hf_sync.py super_squash "${HF_TOKEN}" "${DATASET_ID}"
|
172 |
+
else
|
173 |
+
LAST=$(cat "$SQUASH_FLAG_FILE")
|
174 |
+
DIFF=$((NOW - LAST))
|
175 |
+
if [ $DIFF -ge $SEVEN_DAYS ]; then
|
176 |
+
echo $NOW > "$SQUASH_FLAG_FILE"
|
177 |
+
echo "距离上次合并已超过7天,正在合并历史提交..."
|
178 |
+
python hf_sync.py super_squash "${HF_TOKEN}" "${DATASET_ID}"
|
179 |
+
else
|
180 |
+
echo "距离上次合并未满7天,本次跳过合并历史提交。"
|
181 |
+
fi
|
182 |
+
fi
|
183 |
else
|
184 |
+
echo "错误: 创建备份失败" >&2
|
185 |
fi
|
|
|
186 |
|
187 |
# 清理临时文件
|
188 |
+
rm -f "${temp_backup}"
|
189 |
else
|
190 |
+
echo "警告: 存储目录 ${STORAGE_PATH} 不存在,等待中..." >&2
|
191 |
+
fi
|
192 |
+
|
193 |
+
# 检查 OpenList 服务是否仍在运行
|
194 |
+
if ! kill -0 $OPENLIST_PID 2>/dev/null; then
|
195 |
+
echo "错误: OpenList 服务已停止,退出同步进程" >&2
|
196 |
+
exit 1
|
197 |
fi
|
198 |
|
199 |
# 同步间隔
|
|
|
200 |
echo "下次同步将在 ${SYNC_INTERVAL} 秒后进行..."
|
201 |
sleep $SYNC_INTERVAL
|
202 |
done
|
203 |
}
|
204 |
|
205 |
# 启动同步进程
|
206 |
+
sync_data
|