forked from phoenix/litellm-mirror
(feat) redis-semantic cache on proxy
This commit is contained in:
parent
bdc2091838
commit
2732c47b70
2 changed files with 5 additions and 2 deletions
|
@ -1135,7 +1135,7 @@ class ProxyConfig:
|
||||||
|
|
||||||
verbose_proxy_logger.debug(f"passed cache type={cache_type}")
|
verbose_proxy_logger.debug(f"passed cache type={cache_type}")
|
||||||
|
|
||||||
if cache_type == "redis":
|
if cache_type == "redis" or cache_type == "redis-semantic":
|
||||||
cache_host = litellm.get_secret("REDIS_HOST", None)
|
cache_host = litellm.get_secret("REDIS_HOST", None)
|
||||||
cache_port = litellm.get_secret("REDIS_PORT", None)
|
cache_port = litellm.get_secret("REDIS_PORT", None)
|
||||||
cache_password = litellm.get_secret("REDIS_PASSWORD", None)
|
cache_password = litellm.get_secret("REDIS_PASSWORD", None)
|
||||||
|
@ -1162,6 +1162,9 @@ class ProxyConfig:
|
||||||
f"{blue_color_code}Cache Password:{reset_color_code} {cache_password}"
|
f"{blue_color_code}Cache Password:{reset_color_code} {cache_password}"
|
||||||
)
|
)
|
||||||
print() # noqa
|
print() # noqa
|
||||||
|
if cache_type == "redis-semantic":
|
||||||
|
# by default this should always be async
|
||||||
|
cache_params.update({"redis_semantic_cache_use_async": True})
|
||||||
|
|
||||||
# users can pass os.environ/ variables on the proxy - we should read them from the env
|
# users can pass os.environ/ variables on the proxy - we should read them from the env
|
||||||
for key, value in cache_params.items():
|
for key, value in cache_params.items():
|
||||||
|
@ -4067,7 +4070,6 @@ def _has_user_setup_sso():
|
||||||
async def shutdown_event():
|
async def shutdown_event():
|
||||||
global prisma_client, master_key, user_custom_auth, user_custom_key_generate
|
global prisma_client, master_key, user_custom_auth, user_custom_key_generate
|
||||||
if prisma_client:
|
if prisma_client:
|
||||||
|
|
||||||
verbose_proxy_logger.debug("Disconnecting from Prisma")
|
verbose_proxy_logger.debug("Disconnecting from Prisma")
|
||||||
await prisma_client.disconnect()
|
await prisma_client.disconnect()
|
||||||
|
|
||||||
|
|
|
@ -10,6 +10,7 @@ gunicorn==21.2.0 # server dep
|
||||||
boto3==1.28.58 # aws bedrock/sagemaker calls
|
boto3==1.28.58 # aws bedrock/sagemaker calls
|
||||||
redis==4.6.0 # caching
|
redis==4.6.0 # caching
|
||||||
redisvl==0.0.7 # semantic caching
|
redisvl==0.0.7 # semantic caching
|
||||||
|
numpy==1.24.3 # semantic caching
|
||||||
prisma==0.11.0 # for db
|
prisma==0.11.0 # for db
|
||||||
mangum==0.17.0 # for aws lambda functions
|
mangum==0.17.0 # for aws lambda functions
|
||||||
google-generativeai==0.1.0 # for vertex ai calls
|
google-generativeai==0.1.0 # for vertex ai calls
|
||||||
|
|
Loading…
Add table
Add a link
Reference in a new issue