Spaces:
Running
Running
chore: update something
Browse files
lightweight_embeddings/analytics.py
CHANGED
@@ -6,27 +6,22 @@ from datetime import datetime
|
|
6 |
from collections import defaultdict
|
7 |
from typing import Dict
|
8 |
|
9 |
-
|
10 |
logger = logging.getLogger(__name__)
|
11 |
|
12 |
-
|
13 |
class Analytics:
|
14 |
-
def __init__(self, redis_url: str, sync_interval: int = 60):
|
15 |
"""
|
16 |
Initializes the Analytics class with an async Redis connection and sync interval.
|
17 |
|
18 |
Parameters:
|
19 |
- redis_url: Redis connection URL (e.g., 'redis://localhost:6379/0')
|
20 |
- sync_interval: Interval in seconds for syncing with Redis.
|
|
|
21 |
"""
|
22 |
-
self.
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
socket_connect_timeout=5,
|
27 |
-
retry_on_timeout=True,
|
28 |
-
socket_keepalive=True,
|
29 |
-
)
|
30 |
self.local_buffer = {
|
31 |
"access": defaultdict(
|
32 |
lambda: defaultdict(int)
|
@@ -35,12 +30,24 @@ class Analytics:
|
|
35 |
lambda: defaultdict(int)
|
36 |
), # {period: {model_id: tokens_count}}
|
37 |
}
|
38 |
-
self.sync_interval = sync_interval
|
39 |
self.lock = asyncio.Lock() # Async lock for thread-safe updates
|
40 |
asyncio.create_task(self._start_sync_task())
|
41 |
-
|
42 |
logger.info("Initialized Analytics with Redis connection: %s", redis_url)
|
43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
def _get_period_keys(self) -> tuple:
|
45 |
"""
|
46 |
Returns keys for day, week, month, and year based on the current date.
|
@@ -101,33 +108,93 @@ class Analytics:
|
|
101 |
Synchronizes local buffer data with Redis.
|
102 |
"""
|
103 |
async with self.lock:
|
104 |
-
|
|
|
105 |
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
111 |
|
112 |
-
|
113 |
-
|
114 |
-
|
115 |
-
|
116 |
-
|
117 |
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
122 |
|
123 |
async def _start_sync_task(self):
|
124 |
"""
|
125 |
Starts a background task that periodically syncs data to Redis.
|
|
|
126 |
"""
|
|
|
|
|
127 |
while True:
|
128 |
await asyncio.sleep(self.sync_interval)
|
129 |
try:
|
130 |
await self._sync_to_redis()
|
|
|
131 |
except redis.exceptions.ConnectionError as e:
|
132 |
logger.error("Redis connection error: %s", e)
|
133 |
-
await
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
from collections import defaultdict
|
7 |
from typing import Dict
|
8 |
|
|
|
9 |
logger = logging.getLogger(__name__)
|
10 |
|
|
|
11 |
class Analytics:
|
12 |
+
def __init__(self, redis_url: str, sync_interval: int = 60, max_retries: int = 5):
|
13 |
"""
|
14 |
Initializes the Analytics class with an async Redis connection and sync interval.
|
15 |
|
16 |
Parameters:
|
17 |
- redis_url: Redis connection URL (e.g., 'redis://localhost:6379/0')
|
18 |
- sync_interval: Interval in seconds for syncing with Redis.
|
19 |
+
- max_retries: Maximum number of retries for reconnecting to Redis.
|
20 |
"""
|
21 |
+
self.redis_url = redis_url
|
22 |
+
self.sync_interval = sync_interval
|
23 |
+
self.max_retries = max_retries
|
24 |
+
self.redis_client = self._create_redis_client()
|
|
|
|
|
|
|
|
|
25 |
self.local_buffer = {
|
26 |
"access": defaultdict(
|
27 |
lambda: defaultdict(int)
|
|
|
30 |
lambda: defaultdict(int)
|
31 |
), # {period: {model_id: tokens_count}}
|
32 |
}
|
|
|
33 |
self.lock = asyncio.Lock() # Async lock for thread-safe updates
|
34 |
asyncio.create_task(self._start_sync_task())
|
35 |
+
|
36 |
logger.info("Initialized Analytics with Redis connection: %s", redis_url)
|
37 |
|
38 |
+
def _create_redis_client(self) -> redis.Redis:
|
39 |
+
"""
|
40 |
+
Creates and returns a new Redis client.
|
41 |
+
"""
|
42 |
+
return redis.from_url(
|
43 |
+
self.redis_url,
|
44 |
+
decode_responses=True,
|
45 |
+
health_check_interval=10,
|
46 |
+
socket_connect_timeout=5,
|
47 |
+
retry_on_timeout=True,
|
48 |
+
socket_keepalive=True,
|
49 |
+
)
|
50 |
+
|
51 |
def _get_period_keys(self) -> tuple:
|
52 |
"""
|
53 |
Returns keys for day, week, month, and year based on the current date.
|
|
|
108 |
Synchronizes local buffer data with Redis.
|
109 |
"""
|
110 |
async with self.lock:
|
111 |
+
try:
|
112 |
+
pipeline = self.redis_client.pipeline()
|
113 |
|
114 |
+
# Sync access counts
|
115 |
+
for period, models in self.local_buffer["access"].items():
|
116 |
+
for model_id, count in models.items():
|
117 |
+
redis_key = f"analytics:access:{period}"
|
118 |
+
pipeline.hincrby(redis_key, model_id, count)
|
119 |
|
120 |
+
# Sync token counts
|
121 |
+
for period, models in self.local_buffer["tokens"].items():
|
122 |
+
for model_id, count in models.items():
|
123 |
+
redis_key = f"analytics:tokens:{period}"
|
124 |
+
pipeline.hincrby(redis_key, model_id, count)
|
125 |
|
126 |
+
pipeline.execute()
|
127 |
+
self.local_buffer["access"].clear() # Clear access buffer after sync
|
128 |
+
self.local_buffer["tokens"].clear() # Clear tokens buffer after sync
|
129 |
+
logger.info("Synced analytics data to Redis.")
|
130 |
+
|
131 |
+
except redis.exceptions.ConnectionError as e:
|
132 |
+
logger.error("Redis connection error during sync: %s", e)
|
133 |
+
raise e
|
134 |
+
except Exception as e:
|
135 |
+
logger.error("Unexpected error during Redis sync: %s", e)
|
136 |
+
raise e
|
137 |
|
138 |
async def _start_sync_task(self):
|
139 |
"""
|
140 |
Starts a background task that periodically syncs data to Redis.
|
141 |
+
Implements retry logic with exponential backoff on connection failures.
|
142 |
"""
|
143 |
+
retry_delay = 1 # Initial retry delay in seconds
|
144 |
+
|
145 |
while True:
|
146 |
await asyncio.sleep(self.sync_interval)
|
147 |
try:
|
148 |
await self._sync_to_redis()
|
149 |
+
retry_delay = 1 # Reset retry delay after successful sync
|
150 |
except redis.exceptions.ConnectionError as e:
|
151 |
logger.error("Redis connection error: %s", e)
|
152 |
+
await self._handle_redis_reconnection()
|
153 |
+
except Exception as e:
|
154 |
+
logger.error("Error during sync: %s", e)
|
155 |
+
# Depending on the error, you might want to handle differently
|
156 |
+
|
157 |
+
async def _handle_redis_reconnection(self):
|
158 |
+
"""
|
159 |
+
Handles Redis reconnection with exponential backoff.
|
160 |
+
"""
|
161 |
+
retry_count = 0
|
162 |
+
delay = 1 # Start with 1 second delay
|
163 |
+
|
164 |
+
while retry_count < self.max_retries:
|
165 |
+
try:
|
166 |
+
logger.info("Attempting to reconnect to Redis (Attempt %d)...", retry_count + 1)
|
167 |
+
self.redis_client.close()
|
168 |
+
self.redis_client = self._create_redis_client()
|
169 |
+
# Optionally, perform a simple command to check connection
|
170 |
+
self.redis_client.ping()
|
171 |
+
logger.info("Successfully reconnected to Redis.")
|
172 |
+
return
|
173 |
+
except redis.exceptions.ConnectionError as e:
|
174 |
+
logger.error("Reconnection attempt %d failed: %s", retry_count + 1, e)
|
175 |
+
retry_count += 1
|
176 |
+
await asyncio.sleep(delay)
|
177 |
+
delay *= 2 # Exponential backoff
|
178 |
+
|
179 |
+
logger.critical("Max reconnection attempts reached. Unable to reconnect to Redis.")
|
180 |
+
# Depending on your application's requirements, you might choose to exit or keep retrying indefinitely
|
181 |
+
# For example, to keep retrying:
|
182 |
+
while True:
|
183 |
+
try:
|
184 |
+
logger.info("Retrying to reconnect to Redis...")
|
185 |
+
self.redis_client.close()
|
186 |
+
self.redis_client = self._create_redis_client()
|
187 |
+
self.redis_client.ping()
|
188 |
+
logger.info("Successfully reconnected to Redis.")
|
189 |
+
break
|
190 |
+
except redis.exceptions.ConnectionError as e:
|
191 |
+
logger.error("Reconnection attempt failed: %s", e)
|
192 |
+
await asyncio.sleep(delay)
|
193 |
+
delay = min(delay * 2, 60) # Cap the delay to 60 seconds
|
194 |
+
|
195 |
+
async def close(self):
|
196 |
+
"""
|
197 |
+
Closes the Redis connection gracefully.
|
198 |
+
"""
|
199 |
+
self.redis_client.close()
|
200 |
+
logger.info("Closed Redis connection.")
|
lightweight_embeddings/service.py
CHANGED
@@ -155,7 +155,7 @@ class EmbeddingsService:
|
|
155 |
"""
|
156 |
|
157 |
def __init__(self, config: Optional[ModelConfig] = None):
|
158 |
-
self.lru_cache = LRUCache(maxsize=
|
159 |
|
160 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
161 |
self.config = config or ModelConfig()
|
|
|
155 |
"""
|
156 |
|
157 |
def __init__(self, config: Optional[ModelConfig] = None):
|
158 |
+
self.lru_cache = LRUCache(maxsize=10_000) # Approximate for ~100MB usage
|
159 |
|
160 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
161 |
self.config = config or ModelConfig()
|