OpenSlides/openslides/utils/redis.py

72 lines
2.3 KiB
Python
Raw Normal View History

from typing import Any
from django.conf import settings
from . import logging
logger = logging.getLogger(__name__)
Redis: Wait for replication on writes Since channels_redis does not support dedicated read-redis instances, the autoupdate message may be received before the data was replicated. All workers read the autoupdate message from the write host, so there is a race between getting this message and a finished replication. For large payloads, the replication is slower in the most cases (even more in a distributed setup, where the master and replica are on different nodes). The easy way is to wait for replication. But there is one difficulty: The number of replicas has to be known. There is a new settings-variable "AMOUNT_REPLICAS" which defaults to 1. It needs to be set correctly! If it is too high, every autoupdate will be delayed by 1 seconds because of a timeout witing for non-existent replicas. If it is too low, some autoupdates may be wrong (and not detectable by the client!) becuase of reading from non-synchronised relicas. The other possibility is to fork channel_redis and add the feature of a read-only redis. This ould help, because on a single redis instance all commands are ordered: First, the data is synced, then the autoupdate message. Attention: This means, if redis-replicas are scaled up, one must make sure to read from the same instance. I think this is not possible in the way how dockers overlay networks work. The only way would be to open one connection and reuse the connection from channels_redis in OpenSlides. This would mean a heavy integration of channels_redis (meaning including the source code in our repo). For the first fix, this one is easy and should work.
2020-04-01 09:26:50 +02:00
# Defaults
use_redis = False
use_read_only_redis = False
read_only_redis_amount_replicas = None
read_only_redis_wait_timeout = None
Redis: Wait for replication on writes Since channels_redis does not support dedicated read-redis instances, the autoupdate message may be received before the data was replicated. All workers read the autoupdate message from the write host, so there is a race between getting this message and a finished replication. For large payloads, the replication is slower in the most cases (even more in a distributed setup, where the master and replica are on different nodes). The easy way is to wait for replication. But there is one difficulty: The number of replicas has to be known. There is a new settings-variable "AMOUNT_REPLICAS" which defaults to 1. It needs to be set correctly! If it is too high, every autoupdate will be delayed by 1 seconds because of a timeout witing for non-existent replicas. If it is too low, some autoupdates may be wrong (and not detectable by the client!) becuase of reading from non-synchronised relicas. The other possibility is to fork channel_redis and add the feature of a read-only redis. This ould help, because on a single redis instance all commands are ordered: First, the data is synced, then the autoupdate message. Attention: This means, if redis-replicas are scaled up, one must make sure to read from the same instance. I think this is not possible in the way how dockers overlay networks work. The only way would be to open one connection and reuse the connection from channels_redis in OpenSlides. This would mean a heavy integration of channels_redis (meaning including the source code in our repo). For the first fix, this one is easy and should work.
2020-04-01 09:26:50 +02:00
try:
import aioredis
except ImportError:
Redis: Wait for replication on writes Since channels_redis does not support dedicated read-redis instances, the autoupdate message may be received before the data was replicated. All workers read the autoupdate message from the write host, so there is a race between getting this message and a finished replication. For large payloads, the replication is slower in the most cases (even more in a distributed setup, where the master and replica are on different nodes). The easy way is to wait for replication. But there is one difficulty: The number of replicas has to be known. There is a new settings-variable "AMOUNT_REPLICAS" which defaults to 1. It needs to be set correctly! If it is too high, every autoupdate will be delayed by 1 seconds because of a timeout witing for non-existent replicas. If it is too low, some autoupdates may be wrong (and not detectable by the client!) becuase of reading from non-synchronised relicas. The other possibility is to fork channel_redis and add the feature of a read-only redis. This ould help, because on a single redis instance all commands are ordered: First, the data is synced, then the autoupdate message. Attention: This means, if redis-replicas are scaled up, one must make sure to read from the same instance. I think this is not possible in the way how dockers overlay networks work. The only way would be to open one connection and reuse the connection from channels_redis in OpenSlides. This would mean a heavy integration of channels_redis (meaning including the source code in our repo). For the first fix, this one is easy and should work.
2020-04-01 09:26:50 +02:00
pass
else:
from .redis_connection_pool import ConnectionPool
# set use_redis to true, if there is a value for REDIS_ADDRESS in the settings
2019-01-06 16:22:33 +01:00
redis_address = getattr(settings, "REDIS_ADDRESS", "")
use_redis = bool(redis_address)
if use_redis:
logger.info(f"Redis address {redis_address}")
pool = ConnectionPool({"address": redis_address})
redis_read_only_address = getattr(settings, "REDIS_READ_ONLY_ADDRESS", "")
use_read_only_redis = bool(redis_read_only_address)
if use_read_only_redis:
logger.info(f"Redis read only address {redis_read_only_address}")
read_only_pool = ConnectionPool({"address": redis_read_only_address})
Redis: Wait for replication on writes Since channels_redis does not support dedicated read-redis instances, the autoupdate message may be received before the data was replicated. All workers read the autoupdate message from the write host, so there is a race between getting this message and a finished replication. For large payloads, the replication is slower in the most cases (even more in a distributed setup, where the master and replica are on different nodes). The easy way is to wait for replication. But there is one difficulty: The number of replicas has to be known. There is a new settings-variable "AMOUNT_REPLICAS" which defaults to 1. It needs to be set correctly! If it is too high, every autoupdate will be delayed by 1 seconds because of a timeout witing for non-existent replicas. If it is too low, some autoupdates may be wrong (and not detectable by the client!) becuase of reading from non-synchronised relicas. The other possibility is to fork channel_redis and add the feature of a read-only redis. This ould help, because on a single redis instance all commands are ordered: First, the data is synced, then the autoupdate message. Attention: This means, if redis-replicas are scaled up, one must make sure to read from the same instance. I think this is not possible in the way how dockers overlay networks work. The only way would be to open one connection and reuse the connection from channels_redis in OpenSlides. This would mean a heavy integration of channels_redis (meaning including the source code in our repo). For the first fix, this one is easy and should work.
2020-04-01 09:26:50 +02:00
read_only_redis_amount_replicas = getattr(settings, "AMOUNT_REPLICAS", 1)
logger.info(f"AMOUNT_REPLICAS={read_only_redis_amount_replicas}")
read_only_redis_wait_timeout = getattr(settings, "WAIT_TIMEOUT", 1000)
logger.info(f"WAIT_TIMEOUT={read_only_redis_wait_timeout}")
else:
logger.info("Redis is not configured.")
2019-03-27 22:51:33 +01:00
# TODO: contextlib.asynccontextmanager can be used in python 3.7
class RedisConnectionContextManager:
"""
Async context manager for connections
"""
2019-01-06 16:22:33 +01:00
def __init__(self, read_only: bool) -> None:
self.pool = read_only_pool if read_only and use_read_only_redis else pool
2019-01-06 16:22:33 +01:00
async def __aenter__(self) -> "aioredis.RedisConnection":
self.conn = await self.pool.pop()
return self.conn
async def __aexit__(self, exc_type: Any, exc: Any, tb: Any) -> None:
if exc:
logger.warn(f"Redis Exception: {exc}. Do not reuse connection...")
self.pool.conn_error(self.conn)
else:
self.pool.push(self.conn)
self.conn = None
def get_connection(read_only: bool = False) -> RedisConnectionContextManager:
"""
Returns contextmanager for a redis connection.
"""
return RedisConnectionContextManager(read_only)