timeout connecting to redis hosted on aws ElastiCache

[Jun711]

Hi,

I have a redis server hosted by aws using aws ElastiCache. When I try to connect to it using a lambda and on my local machine, it gives 'timeout connecting to server' error. However, my colleague could connect to this redis using another client in PHP.
I wonder what I did wrong.

try:
        redisClient = redis.Redis(host='***.use1.cache.amazonaws.com', port=6379, db=0, socket_timeout=10)
        print(redisClient)
        try:
            print(redisClient.ping())
            print(redisClient.set('foo','bar'))
            # r.get('foo')
        except Exception as e:
            print('get set error: ', e)
    except Exception as e:
        print('redis err: ', e)

// redisClient Redis<ConnectionPool<Connection<host=***.use1.cache.amazonaws.com,port=6379,db=0>>>

// get set error:  Timeout connecting to server for ping or set

Without setting the socket_timeout parameter, it will time out based on redis-py operation timeout:
Error 60 connecting to ***.use1.cache.amazonaws.com:6379. Operation timed out.

[dejlek]

I have similar problem - our Redis has 20+ thousand clients (long story why is that so) - however the error is 110. I am going to use redis with retry_on_timeout=True and see whether it helps... Here is the exception:

redis.exceptions.ConnectionError: Error 110 connecting to server.example.com:6379. Connection timed out.

[211217613]

When I try to connect to it using a lambda and on my local machine

• You can only connect to elasticache via a machine running in the same region. It's in the docs somewhere.

[grumps]

Yes you need to be in the same region. Also test with the redis-cli too, to rule out firewall rules, routing etc.

[dejlek]

This is happening randomly. Is there are way to tell redis-py to reconnect on connection timeouts with some backoff strategy?

[theodesp]

That's a good PR. Retry with backoff.

[andymccurdy]

@dejlek There's a retry_on_timeout option in redis-py that will attempt to reconnect and reissue a command when a timeout occurs. If the command again times out, the timeout is then raised to the app to handle.

However, be careful with this setting. There is a chance for data duplication or data loss in very specific scenarios. For example, if you send an RPUSH command successfully but time out on reading the reply from RPUSH, retrying the command could cause the data to be duplicated in the list.

[john-feusi-neuro]

I ran into this issue as well, but from a Lambda. For me, there were a few problems that had to be ironed out

• The lambda needs VPC permissions.
• The ElastiCache security group needs an inbound rule from the Lambda security group that allows communication on the Redis port. I thought they could just be in the same security group.
• And the real kicker: I had turned on encryption in-transit. This meant that I needed to pass redis.RedisClietn(... ssl=True). The redis-py page mentions that ssl_cert_reqs needs to be set to None for use with ElastiCache, but that didn't seem to be true in my case. I did however need to pass ssl=True.

It makes sense that ssl=True needed to be set but the connection was just timing out so I went round and round trying to figure out what the problem with the permissions/VPC/SG setup was.

[soumilshah1995]

I am running into similar issue

[github-actions]

This issue is marked stale. It will be closed in 30 days if it is not updated.

[konstankinollc]

Please make sure your Redis placement is in the same availability zone as your server. Multi-AZ will allow you to access your server from multiple AZs.

[omaraltayyan]

Please make sure your Redis placement is in the same availability zone as your server. Multi-AZ will allow you to access your server from multiple AZs.

I am sorry this is not true, Multi-AZ is for increasing the availability of your setup, so if something happens to one of your redis instances the other one will take its place.

You can connect to an ElasticCache instance if it's in the same region, no need to worry about AZs for that.

[samitpal]

Retry with backoff like redis-rb might be useful I guess.