Spaces:
Paused
Paused
more fixes
Browse files- backend/scripts/backup.py +46 -60
backend/scripts/backup.py
CHANGED
@@ -8,18 +8,15 @@ from pathlib import Path
|
|
8 |
from huggingface_hub import HfApi, hf_hub_download, CommitOperationAdd
|
9 |
|
10 |
###############################################################################
|
11 |
-
# 1) Determine
|
12 |
-
# - We read DATA_DIR from the environment, defaulting to "/app/backend/data"
|
13 |
-
# - Then the DB is "webui.db" in that directory.
|
14 |
###############################################################################
|
15 |
DATA_DIR = os.environ.get("DATA_DIR", "/app/backend/data")
|
16 |
DB_FILE_PATH = os.path.join(DATA_DIR, "webui.db")
|
17 |
|
18 |
###############################################################################
|
19 |
-
# 2)
|
20 |
-
# By default, "/tmp" is writable on Hugging Face Spaces.
|
21 |
###############################################################################
|
22 |
-
BACKUP_DIR = "/
|
23 |
TIMESTAMP_FILE_PATH = os.path.join(BACKUP_DIR, "last_backup_time.txt")
|
24 |
DB_GPG_PATH = os.path.join(BACKUP_DIR, "webui.db.gpg")
|
25 |
|
@@ -30,17 +27,14 @@ REPO_DB_GPG_FILE = "db_backup/webui.db.gpg"
|
|
30 |
|
31 |
def ensure_directories():
|
32 |
"""
|
33 |
-
Create
|
34 |
-
We only need to ensure the backup directory is writable, because
|
35 |
-
the DB itself is at DATA_DIR, which might be read-only or read-write
|
36 |
-
depending on your environment.
|
37 |
"""
|
38 |
try:
|
39 |
os.makedirs(BACKUP_DIR, mode=0o755, exist_ok=True)
|
40 |
dir_stat = os.stat(BACKUP_DIR)
|
41 |
-
print(f"Backup directory {BACKUP_DIR}
|
42 |
|
43 |
-
#
|
44 |
test_file = os.path.join(BACKUP_DIR, '.write_test')
|
45 |
with open(test_file, 'w') as f:
|
46 |
f.write('test')
|
@@ -55,7 +49,7 @@ def ensure_directories():
|
|
55 |
|
56 |
def verify_database():
|
57 |
"""
|
58 |
-
|
59 |
"""
|
60 |
if not os.path.exists(DB_FILE_PATH):
|
61 |
print(f"Database file not found at: {DB_FILE_PATH}")
|
@@ -74,33 +68,32 @@ def verify_database():
|
|
74 |
tables = cursor.fetchall()
|
75 |
|
76 |
if result.lower() == "ok" and len(tables) > 0:
|
77 |
-
print("Database integrity verified successfully")
|
78 |
-
print(f"Found {len(tables)} tables in database")
|
79 |
return True
|
80 |
else:
|
81 |
-
print("Database integrity check failed")
|
82 |
if result.lower() != "ok":
|
83 |
print(f"Integrity check result: {result}")
|
84 |
if len(tables) == 0:
|
85 |
-
print("No tables found in database")
|
86 |
return False
|
87 |
except sqlite3.Error as e:
|
88 |
-
print(f"SQLite error
|
89 |
return False
|
90 |
except Exception as e:
|
91 |
-
print(f"Unexpected error
|
92 |
return False
|
93 |
|
94 |
|
95 |
def encrypt_database(passphrase):
|
96 |
"""
|
97 |
-
Encrypt
|
98 |
-
the .gpg file to /tmp/open_webui/db_backup.
|
99 |
"""
|
100 |
try:
|
101 |
print("\nPreparing for database encryption...")
|
102 |
|
103 |
-
#
|
104 |
gnupg_dir = '/root/.gnupg'
|
105 |
os.makedirs(gnupg_dir, mode=0o700, exist_ok=True)
|
106 |
|
@@ -110,6 +103,7 @@ def encrypt_database(passphrase):
|
|
110 |
"--batch",
|
111 |
"--yes",
|
112 |
"--passphrase", passphrase,
|
|
|
113 |
"-c",
|
114 |
"--cipher-algo", "AES256",
|
115 |
"-o", DB_GPG_PATH,
|
@@ -128,42 +122,37 @@ def encrypt_database(passphrase):
|
|
128 |
return False
|
129 |
|
130 |
if os.path.exists(DB_GPG_PATH):
|
131 |
-
|
132 |
-
print(f"Encryption successful. Encrypted file size: {
|
133 |
return True
|
134 |
else:
|
135 |
-
print("GPG reported success but
|
136 |
return False
|
137 |
except Exception as e:
|
138 |
-
print(f"Encryption failed
|
139 |
return False
|
140 |
|
141 |
|
142 |
def get_last_backup_time(repo_id, hf_token):
|
143 |
"""
|
144 |
-
|
145 |
"""
|
146 |
try:
|
147 |
api = HfApi()
|
148 |
-
files = api.list_repo_files(
|
149 |
-
repo_id=repo_id,
|
150 |
-
repo_type="space",
|
151 |
-
token=hf_token
|
152 |
-
)
|
153 |
if REPO_TIMESTAMP_FILE not in files:
|
154 |
-
print("No timestamp file found in repository")
|
155 |
return None
|
156 |
|
157 |
-
|
158 |
repo_id=repo_id,
|
159 |
repo_type="space",
|
160 |
filename=REPO_TIMESTAMP_FILE,
|
161 |
token=hf_token
|
162 |
)
|
163 |
-
with open(
|
164 |
-
|
165 |
-
|
166 |
-
return datetime.datetime.fromisoformat(timestamp_str)
|
167 |
except Exception as e:
|
168 |
print(f"Error getting last backup time: {e}")
|
169 |
return None
|
@@ -171,7 +160,7 @@ def get_last_backup_time(repo_id, hf_token):
|
|
171 |
|
172 |
def save_timestamp_locally():
|
173 |
"""
|
174 |
-
Save the current UTC time to /
|
175 |
"""
|
176 |
try:
|
177 |
now = datetime.datetime.now(datetime.timezone.utc)
|
@@ -190,28 +179,27 @@ def save_timestamp_locally():
|
|
190 |
|
191 |
def backup_db():
|
192 |
"""
|
193 |
-
Main backup
|
194 |
-
|
195 |
-
|
196 |
-
|
197 |
-
|
198 |
-
|
199 |
-
|
200 |
-
|
201 |
"""
|
202 |
passphrase = os.environ.get("BACKUP_PASSPHRASE")
|
203 |
hf_token = os.environ.get("HF_TOKEN")
|
204 |
space_id = os.environ.get("SPACE_ID")
|
205 |
|
206 |
if not all([passphrase, hf_token, space_id]):
|
207 |
-
print("Error: Missing required environment variables
|
208 |
return False
|
209 |
|
210 |
if not ensure_directories():
|
211 |
print("Failed to create or verify backup directories.")
|
212 |
return False
|
213 |
|
214 |
-
# Check threshold
|
215 |
threshold_minutes = int(os.environ.get("BACKUP_THRESHOLD_MINUTES", 120))
|
216 |
if threshold_minutes > 0:
|
217 |
last_backup_dt = get_last_backup_time(space_id, hf_token)
|
@@ -221,28 +209,27 @@ def backup_db():
|
|
221 |
last_backup_dt = last_backup_dt.replace(tzinfo=datetime.timezone.utc)
|
222 |
elapsed = now - last_backup_dt
|
223 |
if elapsed.total_seconds() < threshold_minutes * 60:
|
224 |
-
print(f"Last backup was only {elapsed.total_seconds()/
|
225 |
-
print(f"Threshold is {threshold_minutes} minutes")
|
226 |
-
print("Skipping backup to avoid frequent rebuilds")
|
227 |
return True
|
228 |
else:
|
229 |
-
print("Backup threshold check disabled (
|
230 |
|
231 |
-
# Verify
|
232 |
if not verify_database():
|
233 |
print("Database verification failed, aborting backup.")
|
234 |
return False
|
235 |
|
236 |
-
# Encrypt
|
237 |
if not encrypt_database(passphrase):
|
238 |
print("Database encryption failed, aborting backup.")
|
239 |
return False
|
240 |
|
241 |
-
# Save
|
242 |
if not save_timestamp_locally():
|
243 |
-
print("Warning: Failed to save timestamp
|
244 |
|
245 |
-
# Upload to
|
246 |
print("\nUploading to Hugging Face Spaces...")
|
247 |
try:
|
248 |
api = HfApi()
|
@@ -256,7 +243,6 @@ def backup_db():
|
|
256 |
path_or_fileobj=TIMESTAMP_FILE_PATH
|
257 |
)
|
258 |
]
|
259 |
-
|
260 |
api.create_commit(
|
261 |
repo_id=space_id,
|
262 |
repo_type="space",
|
@@ -267,7 +253,7 @@ def backup_db():
|
|
267 |
print("Backup files uploaded successfully!")
|
268 |
return True
|
269 |
except Exception as e:
|
270 |
-
print(f"Error uploading backup to
|
271 |
return False
|
272 |
|
273 |
|
|
|
8 |
from huggingface_hub import HfApi, hf_hub_download, CommitOperationAdd
|
9 |
|
10 |
###############################################################################
|
11 |
+
# 1) Determine DB location from your env.py (DATA_DIR), defaulting to /app/backend/data
|
|
|
|
|
12 |
###############################################################################
|
13 |
DATA_DIR = os.environ.get("DATA_DIR", "/app/backend/data")
|
14 |
DB_FILE_PATH = os.path.join(DATA_DIR, "webui.db")
|
15 |
|
16 |
###############################################################################
|
17 |
+
# 2) Use /workspace (guaranteed writable on HF Spaces) for backups.
|
|
|
18 |
###############################################################################
|
19 |
+
BACKUP_DIR = "/workspace/open_webui/db_backup"
|
20 |
TIMESTAMP_FILE_PATH = os.path.join(BACKUP_DIR, "last_backup_time.txt")
|
21 |
DB_GPG_PATH = os.path.join(BACKUP_DIR, "webui.db.gpg")
|
22 |
|
|
|
27 |
|
28 |
def ensure_directories():
|
29 |
"""
|
30 |
+
Create/verify /workspace/open_webui/db_backup.
|
|
|
|
|
|
|
31 |
"""
|
32 |
try:
|
33 |
os.makedirs(BACKUP_DIR, mode=0o755, exist_ok=True)
|
34 |
dir_stat = os.stat(BACKUP_DIR)
|
35 |
+
print(f"Backup directory {BACKUP_DIR} exists with permissions: {oct(dir_stat.st_mode)[-3:]}")
|
36 |
|
37 |
+
# Quick test to verify we can write
|
38 |
test_file = os.path.join(BACKUP_DIR, '.write_test')
|
39 |
with open(test_file, 'w') as f:
|
40 |
f.write('test')
|
|
|
49 |
|
50 |
def verify_database():
|
51 |
"""
|
52 |
+
Check that webui.db exists and passes PRAGMA integrity_check.
|
53 |
"""
|
54 |
if not os.path.exists(DB_FILE_PATH):
|
55 |
print(f"Database file not found at: {DB_FILE_PATH}")
|
|
|
68 |
tables = cursor.fetchall()
|
69 |
|
70 |
if result.lower() == "ok" and len(tables) > 0:
|
71 |
+
print("Database integrity verified successfully.")
|
72 |
+
print(f"Found {len(tables)} tables in database.")
|
73 |
return True
|
74 |
else:
|
75 |
+
print("Database integrity check failed.")
|
76 |
if result.lower() != "ok":
|
77 |
print(f"Integrity check result: {result}")
|
78 |
if len(tables) == 0:
|
79 |
+
print("No tables found in database.")
|
80 |
return False
|
81 |
except sqlite3.Error as e:
|
82 |
+
print(f"SQLite error: {e}")
|
83 |
return False
|
84 |
except Exception as e:
|
85 |
+
print(f"Unexpected error: {e}")
|
86 |
return False
|
87 |
|
88 |
|
89 |
def encrypt_database(passphrase):
|
90 |
"""
|
91 |
+
Encrypt /app/backend/data/webui.db to /workspace/open_webui/db_backup/webui.db.gpg.
|
|
|
92 |
"""
|
93 |
try:
|
94 |
print("\nPreparing for database encryption...")
|
95 |
|
96 |
+
# Make sure .gnupg is created with correct permissions
|
97 |
gnupg_dir = '/root/.gnupg'
|
98 |
os.makedirs(gnupg_dir, mode=0o700, exist_ok=True)
|
99 |
|
|
|
103 |
"--batch",
|
104 |
"--yes",
|
105 |
"--passphrase", passphrase,
|
106 |
+
"--pinentry-mode", "loopback", # <--- important in containers
|
107 |
"-c",
|
108 |
"--cipher-algo", "AES256",
|
109 |
"-o", DB_GPG_PATH,
|
|
|
122 |
return False
|
123 |
|
124 |
if os.path.exists(DB_GPG_PATH):
|
125 |
+
size = os.path.getsize(DB_GPG_PATH)
|
126 |
+
print(f"Encryption successful. Encrypted file size: {size:,} bytes")
|
127 |
return True
|
128 |
else:
|
129 |
+
print("GPG reported success, but webui.db.gpg not found.")
|
130 |
return False
|
131 |
except Exception as e:
|
132 |
+
print(f"Encryption failed: {e}")
|
133 |
return False
|
134 |
|
135 |
|
136 |
def get_last_backup_time(repo_id, hf_token):
|
137 |
"""
|
138 |
+
Retrieve last backup timestamp from the HF repo (if it exists).
|
139 |
"""
|
140 |
try:
|
141 |
api = HfApi()
|
142 |
+
files = api.list_repo_files(repo_id=repo_id, repo_type="space", token=hf_token)
|
|
|
|
|
|
|
|
|
143 |
if REPO_TIMESTAMP_FILE not in files:
|
144 |
+
print("No timestamp file found in repository.")
|
145 |
return None
|
146 |
|
147 |
+
tmp_file = hf_hub_download(
|
148 |
repo_id=repo_id,
|
149 |
repo_type="space",
|
150 |
filename=REPO_TIMESTAMP_FILE,
|
151 |
token=hf_token
|
152 |
)
|
153 |
+
with open(tmp_file, "r", encoding="utf-8") as f:
|
154 |
+
stamp_str = f.read().strip()
|
155 |
+
return datetime.datetime.fromisoformat(stamp_str)
|
|
|
156 |
except Exception as e:
|
157 |
print(f"Error getting last backup time: {e}")
|
158 |
return None
|
|
|
160 |
|
161 |
def save_timestamp_locally():
|
162 |
"""
|
163 |
+
Save the current UTC time to /workspace/open_webui/db_backup/last_backup_time.txt
|
164 |
"""
|
165 |
try:
|
166 |
now = datetime.datetime.now(datetime.timezone.utc)
|
|
|
179 |
|
180 |
def backup_db():
|
181 |
"""
|
182 |
+
Main backup process:
|
183 |
+
- Read passphrase, HF creds
|
184 |
+
- Create /workspace/open_webui/db_backup
|
185 |
+
- Possibly skip if threshold not met
|
186 |
+
- Verify DB
|
187 |
+
- Encrypt DB
|
188 |
+
- Save timestamp
|
189 |
+
- Upload to HF
|
190 |
"""
|
191 |
passphrase = os.environ.get("BACKUP_PASSPHRASE")
|
192 |
hf_token = os.environ.get("HF_TOKEN")
|
193 |
space_id = os.environ.get("SPACE_ID")
|
194 |
|
195 |
if not all([passphrase, hf_token, space_id]):
|
196 |
+
print("Error: Missing required environment variables (BACKUP_PASSPHRASE, HF_TOKEN, SPACE_ID).")
|
197 |
return False
|
198 |
|
199 |
if not ensure_directories():
|
200 |
print("Failed to create or verify backup directories.")
|
201 |
return False
|
202 |
|
|
|
203 |
threshold_minutes = int(os.environ.get("BACKUP_THRESHOLD_MINUTES", 120))
|
204 |
if threshold_minutes > 0:
|
205 |
last_backup_dt = get_last_backup_time(space_id, hf_token)
|
|
|
209 |
last_backup_dt = last_backup_dt.replace(tzinfo=datetime.timezone.utc)
|
210 |
elapsed = now - last_backup_dt
|
211 |
if elapsed.total_seconds() < threshold_minutes * 60:
|
212 |
+
print(f"Last backup was only {elapsed.total_seconds()/60:.1f} min ago.")
|
213 |
+
print(f"Threshold is {threshold_minutes} minutes. Skipping backup.")
|
|
|
214 |
return True
|
215 |
else:
|
216 |
+
print("Backup threshold check disabled (0).")
|
217 |
|
218 |
+
# Verify DB
|
219 |
if not verify_database():
|
220 |
print("Database verification failed, aborting backup.")
|
221 |
return False
|
222 |
|
223 |
+
# Encrypt
|
224 |
if not encrypt_database(passphrase):
|
225 |
print("Database encryption failed, aborting backup.")
|
226 |
return False
|
227 |
|
228 |
+
# Save timestamp
|
229 |
if not save_timestamp_locally():
|
230 |
+
print("Warning: Failed to save timestamp, but continuing upload.")
|
231 |
|
232 |
+
# Upload to HF
|
233 |
print("\nUploading to Hugging Face Spaces...")
|
234 |
try:
|
235 |
api = HfApi()
|
|
|
243 |
path_or_fileobj=TIMESTAMP_FILE_PATH
|
244 |
)
|
245 |
]
|
|
|
246 |
api.create_commit(
|
247 |
repo_id=space_id,
|
248 |
repo_type="space",
|
|
|
253 |
print("Backup files uploaded successfully!")
|
254 |
return True
|
255 |
except Exception as e:
|
256 |
+
print(f"Error uploading backup to HF: {e}")
|
257 |
return False
|
258 |
|
259 |
|