05:33:19 GMT <vikas> Hi
10:20:58 GMT <hphuoc25> Hello. I have key-value pairs k:v with TTL 100 and a set S to store keys k's. When one of the key expire, I'd like to remove it from the set S also. How can I do that?
10:22:46 GMT <hphuoc25> Sorry what i meant was, when one of the key expire, I'd like to retrieve its value for further processing. How can I do that?
10:31:28 GMT <minus> hphuoc25: there's keyspace notifications (have to be enabled) that tell you when a key expires. but at that moment the key is already gone
10:31:41 GMT <minus> and you'd need a client listening for that
10:34:31 GMT <hphuoc25> minus: yes I know. When the client receives the key, the value is gone, and I'm not sure how to get the value
10:36:31 GMT <minus> if you want to use redis as timer: create a second key as timer key that expires and don't have the other one expire automatically
10:37:53 GMT <minus> or put timestamps + key names into a sorted set and have a client poll those and expire the keys that are due. this is safer because it also works if the client wasn't connected while a key expired
10:50:59 GMT <hphuoc25> minus: thanks for suggestions :)
12:09:07 GMT <redis946> maxmemory
12:09:13 GMT <redis946> =1
12:09:15 GMT <redis946> exit
12:10:02 GMT <redis156> Hello
13:02:57 GMT <redis156> test
16:46:00 GMT <Guest10483> Question for folks—we've got a cluster set up and are trying to run a script on the slave node, but the slave responds with a redirect back to the master. I've verified that the slot is served by that node, and that the key passed to the script is on the slave. I haven't been able to find similar issues mentioned online. Is this a better fit for the mailing list or a GitHub issue?
17:13:29 GMT <badboy_> Guest10483: if you want to read from a slave in cluster mode you need http://redis.io/commands/readonly
17:26:47 GMT <Guest22041> Hi, we want to use redis clustered over 3 nodes
17:27:13 GMT <Guest22041> what should be the configuration
17:27:45 GMT <Guest22041> our IT is saying that we need to have master node on each node and then 2 slaves to each master
17:28:11 GMT <Guest22041> Is that true or can we just have standard cluster configuration with Master on one node and slaves on other two?
17:28:30 GMT <badboy_> that's not a Redis Cluster then
17:28:36 GMT <badboy_> that's standard replicated Redis
17:29:24 GMT <Guest22041> whcih one? config with One master and two slaves?
17:29:37 GMT <badboy_> yes, that's just replicated Redis
17:30:53 GMT <minus> Guest22041: cluster means to have data partitioned between multiple master instances
17:31:12 GMT <minus> also, long time no see, badboy_
17:31:23 GMT <badboy_> heho
17:31:28 GMT <badboy_> yeah, busy with other stuff
17:31:47 GMT <borei> hi all
17:35:20 GMT <Guest22041> thank you bad boy. I think we misuderstood difference between replicated and clustered and that your suggestion can solve the issue  we see.
17:35:27 GMT <Guest22041> We are trying new configuration as we speak
18:16:38 GMT <borei> need some help with master-slave configuration and failover
18:17:01 GMT <borei> so i have 3 instances of sentinel and 2 instances of redis in the following layout
18:17:18 GMT <borei> (r1, s1) (r2, s2) (s3)
18:17:43 GMT <borei> () - means vm/host where service is running from
18:17:59 GMT <borei> r2 is configured as slave of r1
18:18:33 GMT <borei> when bot redis are up - everything is ok
18:18:57 GMT <borei> r1 is master, r2 slave, haproxy detect it and sends traffic to proper instance
18:19:36 GMT <borei> when im turning off r1, sentinel detects that situation and promotes r2 to master
18:20:42 GMT <borei> but approx 5 seconds later it demotes r2 to slave and point master to ip of r2
18:20:48 GMT <borei> that is confusing me
18:21:09 GMT <borei> eventually both nodes has role:slave and all traffic stopped
18:21:30 GMT <borei> apperantly im missing something simple, but can't find
18:22:53 GMT <borei> any ideas ?
18:41:09 GMT <via> why is it that a scan with a pattern that matches nothign would take 5 seconds whereas the same pattern with keys only takes <100 ms?
18:43:30 GMT <via> nevermind, i'm an idiot
18:52:24 GMT <borei> anybody ?
19:09:58 GMT <minus> via: what was the issue?
19:11:07 GMT <via> i was timing python's scan_iter, which was including all the round trips at a default coutn of 10
19:11:58 GMT <minus> it was doing multiple calls to scan until it was through the entire keyspace because it didn't find any keys?
19:12:31 GMT <via> yes, and in my head i thought scan's count applied to results, not the search itself
19:12:34 GMT <via> was wrong, and it works now
19:14:50 GMT <borei> anybody can help with master-slave configuration ?
19:17:36 GMT <minus> sorry, borei, no idea what could be going wrong there
19:18:05 GMT <borei> any known bugs ? im on 3.0.6 version
19:18:43 GMT <tgodar> so to use redis as a drop in replacement for memcached... each node needs to talk to the other on a second port?  are there other things I am overlooking?
19:20:23 GMT <minus> what's "each node"?
19:21:16 GMT <tgodar> instance?
19:22:56 GMT <tgodar> let me rephrase this, to use redis as a drop in replacement for memcached what all do I need to do?  i.e. think I need to configure it to not use persistence
19:23:27 GMT <tgodar> I have multiple frontend web servers I would like to run redis on as I did memcached to get a distributed cache
19:24:06 GMT <tgodar> memcached uses a consistent hashing approach - seems redis does not...
19:24:25 GMT <tgodar> I am trying to understand what is entailed.  thanks
19:25:19 GMT <mandy> Hi my rediskey is deleting automatically
19:26:04 GMT <mandy> i have records to set and read the value set next iteration
19:26:20 GMT <mandy> but strange HEXIST returning me
19:26:25 GMT <mandy> error
19:26:33 GMT <mandy> no key exists
19:29:25 GMT <mandy> any help
19:29:29 GMT <minus> tgodar: if you want partitioning you need redis cluster; it does use consistent hashing (or something similar at least)
19:30:03 GMT <mandy> no
19:30:13 GMT <minus> mandy: your key expires? then obviously you can't query it afterwards
19:30:31 GMT <tgodar> minus: is redis cluster something different or just how redis happens to be configured?
19:30:32 GMT <mandy> i have not put any expiry options
19:30:54 GMT <minus> tgodar: different configuration
19:30:59 GMT <minus> tgodar:
19:31:02 GMT <minus> http://redis.io/topics/cluster-tutorial
19:32:38 GMT <tgodar> ok, so my cache size is what is alloted on a single server, and it replicates everything across instances :(
19:33:23 GMT <tgodar> or, wait, I'm reading the wrong thing
19:33:37 GMT <minus> yes, don't confuse cluster with sentinel
19:33:46 GMT <minus> sentinel is HA-only
19:34:15 GMT <minus> there's also plain master/slave replication
19:34:39 GMT <tgodar> yeah, ok, this might work then, thanks minus.  lot to gleen here... do you happen to know if the redis configuraiton needs to be aware of other servers or does the client (phpredis or what not) configuration only need to know that?
19:35:06 GMT <minus> sorry, haven't used cluster yet
19:35:18 GMT <tgodar> that second port being open makes me thing the redis (server) config needs to be made aware...
19:36:02 GMT <minus> the servers need to be aware of each other of course
19:36:47 GMT <minus> for the client i don't know how you get the list of servers
19:37:39 GMT <minus> ah, you just need one server of the cluster, then the client can discover the rest, nice
19:38:00 GMT <minus> i should spin up one of those some time
19:40:24 GMT <borei> seems like it's getting crayzy if bind option is not explicit
19:41:02 GMT <minus> what do you mean, no bind at all?
19:41:40 GMT <borei> by default bind is commented out, it means that redis is listening *:6379
19:41:56 GMT <borei> including 127.0.0.1
19:42:18 GMT <minus> oh, maybe your redis is running in protected mode and that fucks shit up?
19:42:40 GMT <borei> what does it mean protected mode ?
19:43:59 GMT <tgodar> minus thanks, was thinking about implications for automation when adding dropping servers instnaces
19:45:44 GMT <minus> borei: https://www.reddit.com/r/redis/comments/3zv85m/new_security_feature_redis_protected_mode/
19:47:19 GMT <borei> ic
19:58:55 GMT <wilornel__> We have a group project at school where we "Reverse Engineer" an open-source project. We look through it and the final paper is an explanation of how oyu would build it, an explanation of the software architecture. We just decided on Redis. I was wondering, is there some part of the docs that you wish could be improved?
19:59:02 GMT <wilornel__> (let me know if my message cut off please)
19:59:31 GMT <borei> minus: thanks for heads-up !
20:00:14 GMT <borei> nothing in doc about sentinel is saying about protected mode
21:41:37 GMT <Whisket> Is it the time the cmd was received? the time is was processed? some meaningless timestamp?
22:44:51 GMT <Whisket> What is the source of the timestamp for `redis monitor`? Running the command I'll see multiple lines appear all at once with timestamps ranging from 30 seconds ago to now
22:45:01 GMT <Whisket> Is it the time the cmd was received? the time is was processed? some meaningless timestamp?
23:25:34 GMT <Whisket> Anybody?
23:25:47 GMT <Whisket> Anybody at all know what that timestamp is?