Spaces:
Paused
Paused
Update sync_manager.py
Browse files- sync_manager.py +44 -129
sync_manager.py
CHANGED
|
@@ -1,195 +1,110 @@
|
|
| 1 |
#!/usr/bin/env python3
|
| 2 |
-
"""
|
| 3 |
-
Sync Manager - Automatically backs up PostgreSQL to HuggingFace Xet Dataset
|
| 4 |
-
"""
|
| 5 |
|
| 6 |
import os
|
| 7 |
import sys
|
| 8 |
import time
|
| 9 |
import subprocess
|
| 10 |
-
import tempfile
|
| 11 |
import signal
|
| 12 |
from datetime import datetime
|
| 13 |
from pathlib import Path
|
| 14 |
|
| 15 |
-
|
| 16 |
-
SYNC_INTERVAL = int(os.environ.get('SYNC_INTERVAL', 300)) # 5 minutes
|
| 17 |
HF_TOKEN = os.environ.get('HF_TOKEN', '')
|
| 18 |
-
XET_DATASET = os.environ.get('XET_DATASET', '
|
| 19 |
-
POSTGRES_HOST = 'localhost'
|
| 20 |
-
POSTGRES_PORT = 5432
|
| 21 |
-
POSTGRES_USER = 'postgres'
|
| 22 |
-
POSTGRES_DB = 'appdb'
|
| 23 |
-
PGPASSWORD = os.environ.get('POSTGRES_PASSWORD', 'postgres123')
|
| 24 |
|
| 25 |
-
# Global flag for graceful shutdown
|
| 26 |
running = True
|
| 27 |
|
|
|
|
|
|
|
| 28 |
|
| 29 |
-
def
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
import psycopg2
|
| 38 |
-
|
| 39 |
-
for i in range(max_retries):
|
| 40 |
-
try:
|
| 41 |
-
conn = psycopg2.connect(
|
| 42 |
-
host=POSTGRES_HOST,
|
| 43 |
-
port=POSTGRES_PORT,
|
| 44 |
-
user=POSTGRES_USER,
|
| 45 |
-
password=PGPASSWORD,
|
| 46 |
-
database='postgres'
|
| 47 |
-
)
|
| 48 |
-
conn.close()
|
| 49 |
-
log("β
PostgreSQL is ready")
|
| 50 |
return True
|
| 51 |
-
|
| 52 |
-
log(f"β³ Waiting for PostgreSQL... ({i+1}/{max_retries})")
|
| 53 |
-
time.sleep(2)
|
| 54 |
-
|
| 55 |
-
log("β PostgreSQL did not become ready")
|
| 56 |
return False
|
| 57 |
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
|
|
|
|
|
|
| 61 |
try:
|
| 62 |
backup_dir = Path('/data/backup')
|
| 63 |
backup_dir.mkdir(parents=True, exist_ok=True)
|
| 64 |
-
|
| 65 |
backup_file = backup_dir / 'pg_backup.sql'
|
| 66 |
|
| 67 |
-
env = os.environ.copy()
|
| 68 |
-
env['PGPASSWORD'] = PGPASSWORD
|
| 69 |
-
|
| 70 |
# Create backup
|
| 71 |
result = subprocess.run(
|
| 72 |
-
[
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
'-p', str(POSTGRES_PORT),
|
| 76 |
-
'-U', POSTGRES_USER,
|
| 77 |
-
'-d', POSTGRES_DB,
|
| 78 |
-
'--clean',
|
| 79 |
-
'--if-exists',
|
| 80 |
-
'-f', str(backup_file)
|
| 81 |
-
],
|
| 82 |
-
env=env,
|
| 83 |
-
capture_output=True,
|
| 84 |
-
text=True
|
| 85 |
)
|
| 86 |
|
| 87 |
-
if result.returncode =
|
| 88 |
-
size = backup_file.stat().st_size
|
| 89 |
-
log(f"β
Backup created: {backup_file} ({size} bytes)")
|
| 90 |
-
return str(backup_file)
|
| 91 |
-
else:
|
| 92 |
log(f"β Backup failed: {result.stderr}")
|
| 93 |
-
return
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
def upload_to_xet(backup_file: str) -> bool:
|
| 101 |
-
"""Upload backup to HuggingFace Xet dataset."""
|
| 102 |
-
if not HF_TOKEN:
|
| 103 |
-
log("β οΈ No HF_TOKEN set, skipping upload")
|
| 104 |
-
return False
|
| 105 |
-
|
| 106 |
-
try:
|
| 107 |
from huggingface_hub import HfApi, create_repo
|
| 108 |
|
| 109 |
api = HfApi(token=HF_TOKEN)
|
| 110 |
|
| 111 |
-
# Create dataset repo if it doesn't exist
|
| 112 |
try:
|
| 113 |
-
create_repo(
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
|
| 117 |
-
token=HF_TOKEN,
|
| 118 |
-
exist_ok=True
|
| 119 |
-
)
|
| 120 |
-
except Exception as e:
|
| 121 |
-
log(f"βΉοΈ Repo creation note: {e}")
|
| 122 |
|
| 123 |
-
# Upload backup file
|
| 124 |
api.upload_file(
|
| 125 |
-
path_or_fileobj=backup_file,
|
| 126 |
path_in_repo="backup/pg_backup.sql",
|
| 127 |
repo_id=XET_DATASET,
|
| 128 |
repo_type="dataset",
|
| 129 |
token=HF_TOKEN,
|
| 130 |
-
commit_message=f"
|
| 131 |
)
|
| 132 |
|
| 133 |
-
log(f"β
|
| 134 |
-
return True
|
| 135 |
|
| 136 |
except Exception as e:
|
| 137 |
-
log(f"β
|
| 138 |
-
return False
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
def sync_once():
|
| 142 |
-
"""Perform one sync cycle."""
|
| 143 |
-
log("π Starting sync cycle...")
|
| 144 |
-
|
| 145 |
-
backup_file = create_backup()
|
| 146 |
-
if backup_file:
|
| 147 |
-
upload_to_xet(backup_file)
|
| 148 |
-
|
| 149 |
-
log("β
Sync cycle complete")
|
| 150 |
-
|
| 151 |
|
| 152 |
def signal_handler(signum, frame):
|
| 153 |
-
"""Handle shutdown signals."""
|
| 154 |
global running
|
| 155 |
-
log("π
|
| 156 |
running = False
|
| 157 |
-
|
| 158 |
-
log("π Sync manager shutting down")
|
| 159 |
sys.exit(0)
|
| 160 |
|
| 161 |
-
|
| 162 |
def main():
|
| 163 |
-
"
|
| 164 |
-
log("
|
| 165 |
-
log(f"
|
| 166 |
-
log(f"β° Sync interval: {SYNC_INTERVAL} seconds")
|
| 167 |
|
| 168 |
-
# Set up signal handlers
|
| 169 |
signal.signal(signal.SIGTERM, signal_handler)
|
| 170 |
signal.signal(signal.SIGINT, signal_handler)
|
| 171 |
|
| 172 |
-
# Wait for PostgreSQL
|
| 173 |
if not wait_for_postgres():
|
| 174 |
-
log("β
|
| 175 |
return
|
| 176 |
|
| 177 |
-
# Initial delay
|
| 178 |
-
time.sleep(30)
|
| 179 |
|
| 180 |
-
# Main loop
|
| 181 |
while running:
|
| 182 |
-
|
| 183 |
-
sync_once()
|
| 184 |
-
except Exception as e:
|
| 185 |
-
log(f"β Sync error: {e}")
|
| 186 |
-
|
| 187 |
-
# Sleep with interrupt handling
|
| 188 |
for _ in range(SYNC_INTERVAL):
|
| 189 |
if not running:
|
| 190 |
break
|
| 191 |
time.sleep(1)
|
| 192 |
|
| 193 |
-
|
| 194 |
if __name__ == '__main__':
|
| 195 |
main()
|
|
|
|
| 1 |
#!/usr/bin/env python3
|
| 2 |
+
"""Sync PostgreSQL backups to HuggingFace Xet Dataset"""
|
|
|
|
|
|
|
| 3 |
|
| 4 |
import os
|
| 5 |
import sys
|
| 6 |
import time
|
| 7 |
import subprocess
|
|
|
|
| 8 |
import signal
|
| 9 |
from datetime import datetime
|
| 10 |
from pathlib import Path
|
| 11 |
|
| 12 |
+
SYNC_INTERVAL = int(os.environ.get('SYNC_INTERVAL', 300))
|
|
|
|
| 13 |
HF_TOKEN = os.environ.get('HF_TOKEN', '')
|
| 14 |
+
XET_DATASET = os.environ.get('XET_DATASET', '')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
|
|
|
| 16 |
running = True
|
| 17 |
|
| 18 |
+
def log(msg):
|
| 19 |
+
print(f"[{datetime.now().strftime('%H:%M:%S')}] {msg}", flush=True)
|
| 20 |
|
| 21 |
+
def wait_for_postgres(retries=30):
|
| 22 |
+
for i in range(retries):
|
| 23 |
+
result = subprocess.run(
|
| 24 |
+
['pg_isready', '-h', 'localhost', '-p', '5432'],
|
| 25 |
+
capture_output=True
|
| 26 |
+
)
|
| 27 |
+
if result.returncode == 0:
|
| 28 |
+
log("β
PostgreSQL ready")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
return True
|
| 30 |
+
time.sleep(2)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31 |
return False
|
| 32 |
|
| 33 |
+
def backup_and_upload():
|
| 34 |
+
if not HF_TOKEN or not XET_DATASET:
|
| 35 |
+
log("β οΈ HF_TOKEN or XET_DATASET not set")
|
| 36 |
+
return
|
| 37 |
+
|
| 38 |
try:
|
| 39 |
backup_dir = Path('/data/backup')
|
| 40 |
backup_dir.mkdir(parents=True, exist_ok=True)
|
|
|
|
| 41 |
backup_file = backup_dir / 'pg_backup.sql'
|
| 42 |
|
|
|
|
|
|
|
|
|
|
| 43 |
# Create backup
|
| 44 |
result = subprocess.run(
|
| 45 |
+
['pg_dump', '-h', 'localhost', '-U', 'postgres', '-d', 'appdb',
|
| 46 |
+
'--clean', '--if-exists', '-f', str(backup_file)],
|
| 47 |
+
capture_output=True, text=True
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
)
|
| 49 |
|
| 50 |
+
if result.returncode != 0:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
log(f"β Backup failed: {result.stderr}")
|
| 52 |
+
return
|
| 53 |
+
|
| 54 |
+
log(f"β
Backup created ({backup_file.stat().st_size} bytes)")
|
| 55 |
+
|
| 56 |
+
# Upload to HuggingFace
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
from huggingface_hub import HfApi, create_repo
|
| 58 |
|
| 59 |
api = HfApi(token=HF_TOKEN)
|
| 60 |
|
|
|
|
| 61 |
try:
|
| 62 |
+
create_repo(XET_DATASET, repo_type="dataset", private=True,
|
| 63 |
+
token=HF_TOKEN, exist_ok=True)
|
| 64 |
+
except:
|
| 65 |
+
pass
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
|
|
|
|
| 67 |
api.upload_file(
|
| 68 |
+
path_or_fileobj=str(backup_file),
|
| 69 |
path_in_repo="backup/pg_backup.sql",
|
| 70 |
repo_id=XET_DATASET,
|
| 71 |
repo_type="dataset",
|
| 72 |
token=HF_TOKEN,
|
| 73 |
+
commit_message=f"Backup {datetime.now().isoformat()}"
|
| 74 |
)
|
| 75 |
|
| 76 |
+
log(f"β
Uploaded to {XET_DATASET}")
|
|
|
|
| 77 |
|
| 78 |
except Exception as e:
|
| 79 |
+
log(f"β Error: {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 80 |
|
| 81 |
def signal_handler(signum, frame):
|
|
|
|
| 82 |
global running
|
| 83 |
+
log("π Shutting down, final backup...")
|
| 84 |
running = False
|
| 85 |
+
backup_and_upload()
|
|
|
|
| 86 |
sys.exit(0)
|
| 87 |
|
|
|
|
| 88 |
def main():
|
| 89 |
+
log("π Sync Manager started")
|
| 90 |
+
log(f"π¦ Dataset: {XET_DATASET or 'NOT SET'}")
|
| 91 |
+
log(f"β° Interval: {SYNC_INTERVAL}s")
|
|
|
|
| 92 |
|
|
|
|
| 93 |
signal.signal(signal.SIGTERM, signal_handler)
|
| 94 |
signal.signal(signal.SIGINT, signal_handler)
|
| 95 |
|
|
|
|
| 96 |
if not wait_for_postgres():
|
| 97 |
+
log("β PostgreSQL not available")
|
| 98 |
return
|
| 99 |
|
| 100 |
+
time.sleep(30) # Initial delay
|
|
|
|
| 101 |
|
|
|
|
| 102 |
while running:
|
| 103 |
+
backup_and_upload()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
for _ in range(SYNC_INTERVAL):
|
| 105 |
if not running:
|
| 106 |
break
|
| 107 |
time.sleep(1)
|
| 108 |
|
|
|
|
| 109 |
if __name__ == '__main__':
|
| 110 |
main()
|