02:05:20 GMT <braeden> howdy
02:27:29 GMT <jack_rabbit> Thought experiment with Redis: I have a set of string that may be updated frequently. I want to do string searches on this set.
02:28:05 GMT <jack_rabbit> As far as I can tell, there's no quick way to do this on sets, so my idea is to keep a sorted list of the contents of the set which is generated every so often so I can do a binary search - the search doesn't need to be totally up-to-date.
02:29:37 GMT <jack_rabbit> My question about this is whether or not there's a way to do a quick and atomic set on the key holding the sorted list, so I can generate the list in the background and then swap it into the "live" key.
02:34:14 GMT <jack_rabbit> Does this make any sense as a scheme?
12:24:45 GMT <repys> how can I test that redis is working properly in master slave?
12:26:12 GMT <repys> ?
12:39:23 GMT <Habbie> think of a piece of data
12:39:26 GMT <Habbie> check that the slave does not have it
12:39:28 GMT <Habbie> write it to the master
12:39:31 GMT <Habbie> check that the slave now has it
15:37:46 GMT <repys> I am starting redis but it is in listen on the tcp6 port why that?
15:37:53 GMT <repys> how can I put redis in listening on tcp4?
16:00:16 GMT <minus> configure the bind directive
18:58:05 GMT <cookie913> I am looking at this: http://shokunin.co/blog/2014/11/11/operational_redis.html specifically this variable tcp-backlog
18:58:14 GMT <cookie913> it recommends: tcp-backlog 65536
18:58:27 GMT <cookie913> and it says: you will need this to be higher if you have many connections
18:59:24 GMT <cookie913> how do I know that I have too many connections? I have 4000 connections opened right now and my tcp-backlog is still set to the default of 511
18:59:41 GMT <cookie913> or is there some info / metrics that will tell me that my tcp-backlog is too small?
18:59:56 GMT <cookie913> I am using redis 3.0.6
18:59:57 GMT <minus> it's the length of the waiting queue for connections that are incoming but have not been accepted
19:00:25 GMT <cookie913> but how do I measure that… basically I like to know whether my tcp-backlog is too small...
19:00:28 GMT <minus> unless you have a lot of new connections constantly it doesn't matter
19:00:36 GMT <minus> s/constantly/at once/
19:01:14 GMT <minus> if you establish 1000 connections concurrently some of them may fail because the backlog is full
19:01:16 GMT <cookie913> I was looking at the source code, there's no clear indication if that is too small… I mean it gives errors if sockets are not created… is that an indication that the tcp-backlog is too small?
19:03:36 GMT <cookie913> I see… so I won't see errors on the server side? But on the client side, I should see some errors like 'Cannot connect to redis server'?
19:03:51 GMT <minus> yeah
19:04:24 GMT <cookie913> thanks
19:05:05 GMT <minus> you could try to figure out what error you'd get. simply open a tcp socket and listen on it with a backlog of 1, e.g. with python, then connect to it with netcat and just never accept the connections