00:51:44 GMT <SITM> hello -- is there a fork of Redis that can run with zram or compressed LZ4 memory?
10:51:15 GMT <badboy_> SITM: isn't zram fully transparent?
12:20:21 GMT <dudeji> hi , i am not able to find out redis-cli file inside src folder ? how can i start redis-server from source code ?
12:40:23 GMT <naquad> hi
12:40:48 GMT <naquad> what should i pick: first level key like "xyz_{id}" or hash "xyz" with key = id?
12:41:27 GMT <minus> i have a great answer for you: depends
13:06:47 GMT <naquad> minus, well, here's some context: about 2-3k keys, object size is < 1kb (encoded json)
13:07:45 GMT <badboy_> naquad: it still depends
13:07:50 GMT <naquad> on what?
13:07:56 GMT <badboy_> it comes down to how you want to access your data
13:08:08 GMT <naquad> how ofter or ... ?
13:08:47 GMT <badboy_> do you always know the id? do you always need only one object?
13:09:14 GMT <badboy_> there were some experiments where putting it in hashes saves some memory
13:09:19 GMT <badboy_> but you would need to test that for yourself
13:14:22 GMT <naquad> badboy_, i always know id, i always need only one object
13:26:37 GMT <minus> then xzy:{id} is the primary choice, and as badboy_ said hashes might help to save space (but only for a fix amount of them, refer the optimization manual)
13:26:42 GMT <minus> ^ naquad
13:26:48 GMT <naquad> thanks
15:59:14 GMT <cabezza> hi, I've the following situation, I'm trying the subscribe to TTL notifications but I don't only what the notification but the actual key,value that have expired. is that possible? I haven't seen any mention of the being possible at https://redis.io/topics/notifications does anyone have done anything like this?
16:13:28 GMT <minus> cabezza: you do get the key in the notification but there is no way to get the value
16:14:21 GMT <cabezza> minus: I was afraid of that... thnks for your answer
18:43:22 GMT <FatDarrel> Hello everyone
18:44:36 GMT <FatDarrel> does anyone know how to set expiration on all the keys to a certain time ?
18:44:48 GMT <FatDarrel> expire all keys at 10am?
18:53:41 GMT <badboy_> go through all of them and expireata ?
18:53:43 GMT <badboy_> *expireat
18:54:46 GMT <badboy_> There's no single command to do that
19:56:09 GMT <crayon> when using consul with redis, does this also support client side cert validation?
19:56:28 GMT <crayon> even at the API layer
19:58:53 GMT <badboy_> Redis does not, I never used Consul, so I don't even know how those two technologies would interact with each other
19:59:25 GMT <crayon> so the only way to achieve encryption at the API is spipe?
20:00:18 GMT <badboy_> in talking to Redis? yes, it needs an external service for encryption
20:00:32 GMT <badboy_> (vpn, ssh, spipe, stunnel, ...)
20:08:51 GMT <crayon> hrm...
21:39:16 GMT <matt_> hi, i'm looking for a way to keep state information external from an app which would mostly consist of writes and needs to be resilient
21:39:29 GMT <matt_> would redis be a good choice for that?
21:39:45 GMT <minus> define resilient
21:39:53 GMT <matt_> the only info I seem to be able to find on clustering only states a single write master
21:40:12 GMT <matt_> so I can loose any redis node and the clients will be able to jump to another and continue writing their state
21:40:34 GMT <minus> clustering paritions the data between multiple instances, but there is no master-master replication
21:41:11 GMT <minus> both, cluster and sentinel, provide failover by promoting a slave to master
21:41:15 GMT <matt_> ok, I dont need multi master just the ability to have a slave node become a master when the initial master disappears
21:41:21 GMT <matt_> ahh ok
21:42:58 GMT <matt_> cool, so in the config the 'slaveof' statement is that just so the host can initially find the current master or another node for discovery?
21:43:25 GMT <minus> yes
21:43:28 GMT <matt_> ok cool
21:44:17 GMT <minus> keep in mind that redis is not ACID - incurring a node failure can very well lose a few writes
21:45:06 GMT <matt_> yea thats ok, most of the time processes will be over writing the same values, its just so they know where they were if they are killed or the host they are on dies
21:45:49 GMT <matt_> so when they are restarted they can continue, if they replay a couple of things its not the end of the world, also they would have to fail at exactly the same time as the redis node
21:46:16 GMT <matt_> I can stop them being on the same physical node from config on the scheduler
22:13:13 GMT <badboy_> matt_: Redis on its own does not do the failover though!
22:13:17 GMT <badboy_> you need sentinel to do that
22:15:26 GMT <matt_> badboy_: ok, where does that sit? on the same host as the client app?
22:15:33 GMT <matt_> reverse proxy type of thing?
22:19:53 GMT <badboy_> you probably want 3 instances of Sentinel as well
22:19:57 GMT <badboy_> (at least 3)
22:20:02 GMT <badboy_> where to place them is up to you
22:20:16 GMT <badboy_> read up on: https://redis.io/topics/sentinel
22:20:25 GMT <badboy_> also: http://fnordig.de/2015/06/01/redis-sentinel-and-redis-cluster/
22:21:06 GMT <matt_> badboy_: ok cheers
22:21:13 GMT <matt_> i got the redis nodes up :)
22:24:59 GMT <matt_> humm kind of sounds like sentinel isn't a reverse proxy but an 'watching from the side' type of setup
22:28:11 GMT <badboy_> yes
22:43:06 GMT <crayon> hm, any way to achieve encryption in redis for free?