07:10:03 GMT <samof76> i have a master slave setup(3.2.4)
07:10:40 GMT <samof76> and three sentinels watching this setup, I have brought down the master, but the sentinels are not promoting slave to master
07:10:43 GMT <samof76> :(
07:11:33 GMT <samof76> https://gist.github.com/samof76/bc42304d258395b8f95e8c67c2d0bdcf
08:28:57 GMT <spetsnax> I have a list of airports with their geo cordinates (lat & long) in my hand. Now I want to get user's geo position and search through nearby airports to this coordinates. Can I do this with REDIS ?
14:27:20 GMT <max06> Good morning. I'm running a small redis cluster with 3 Masters and 6 Slaves, each of them on a separate virtual machine in the cloud. For upgrade purposes I have to do a rolling restart of every single machine.
14:28:00 GMT <max06> After a restart, I noticed that one of the slaves selected a new master, so I have one master with 2 replicas, one with one replica, and one with 3 replicas.
14:28:16 GMT <max06> I wonder how this could happen?
16:48:02 GMT <miha-_> hello! Have quite a big amount of data. Around 125000000 of entries. Key is like 1231231231 and value like 10. The issue is that this takes aroung 6gb of space for mysql. I am trying to put this in redis but I have issue with memory. Now I have assigned aroung 15gb of mem for this VM. How can this be optimized?
16:50:26 GMT <zetab> why redis? why not elasticsearch?
16:50:34 GMT <zetab> what kind of data that is?
16:51:27 GMT <miha-_> zetab: hello. Just telephone numbers, i need to do quick look up, as this goes for voip
16:52:14 GMT <miha-_> zetab: this voip sw has api for redis, that is why :)
16:52:42 GMT <zetab> well hopefully you know that redis might lose all data in reboot?
16:52:56 GMT <zetab> some guys are using it as "database" but well
16:53:31 GMT <miha-_> zetab: it is not persistent? I tried with restart and it looks like data is preserved
16:54:04 GMT <miha-_> zetab: what would u suggest
16:54:07 GMT <minus> it does persist to disk when configured to
16:54:14 GMT <zetab> yes when you shutdown it in controlled manner it will save stuff to disk
16:54:23 GMT <minus> though it's not ACID or anything like that
16:54:24 GMT <zetab> but what will happen if someone just takes power cable from computer?
16:54:37 GMT <minus> it periodically saves, zetab
16:55:19 GMT <zetab> k
16:55:23 GMT <minus> what's the access pattern on the data, miha-_?
16:55:26 GMT <miha-_> not an issue. this is done only one time
16:55:32 GMT <miha-_>  zfdxdfzjii
16:55:38 GMT <miha-_> sorry
16:55:41 GMT <minus> ;D
16:56:01 GMT <miha-_> access pattern, what do u mean by that?
16:56:19 GMT <minus> how do you read from/write to it
16:56:27 GMT <minus> is it static data?
16:56:34 GMT <miha-_> minus: yes, static data
16:56:41 GMT <zetab> ok then that is not problem
16:56:42 GMT <minus> is it just a key-value lookup?
16:57:12 GMT <miha-_> maybe something is add, but this is maybe on time per month, only one key value inout
16:57:15 GMT <miha-_> inout*
16:57:21 GMT <miha-_> minus: yes :)
16:57:30 GMT <miha-_> i just need to optimze this
16:57:50 GMT <zetab> what about redis cluster?
16:58:17 GMT <zetab> then you could maybe share that load between servers
16:58:23 GMT <zetab> load = entries
16:58:34 GMT <miha-_> zetab: ok, but first i need to get data in one server :)
16:58:42 GMT <miha-_> how to optimeze this :)
16:58:58 GMT <zetab> i have not worked with that big datasets :D
16:58:59 GMT <zetab> no idea
16:59:04 GMT <minus> what is it resolving, phone numbers to something or the other way around?
17:00:01 GMT <miha-_> yes, it looks for phonenumber and I must back some id, which belongs to this number
17:00:16 GMT <minus> is your software's redis access very static or scriptable? if it's just a "get key with prefix xyz:" then you probably can't do much
17:01:11 GMT <miha-_> minus: it can be scribtable
17:01:36 GMT <minus> for that type of data i'd just drop the data into a file in binary format (maybe strings for phone numbers) and just write a simple program to run a binary search on it, should be about the fastest
17:01:53 GMT <minus> okay, then you can try putting the phone numbers into hashes
17:02:24 GMT <minus> if your number is 0123456789 then you go like `HSET 012345 6789 id123`
17:03:04 GMT <minus> and set hash-max-ziplist-entries to 10000
17:03:22 GMT <minus> that should reduce memory a bunch
17:03:51 GMT <miha-_> minus: so if my number is like 12345678, i must do it like hset 1234 5678 10?
17:04:07 GMT <miha-_> minus: i must not do like 12345678 12345678 10
17:04:08 GMT <miha-_> ?
17:04:56 GMT <minus> for the lookup you'll have to do a HGET 012345 6789 then
17:05:18 GMT <minus> you basically divide the dataset into buckets of 10000 numbers
17:06:53 GMT <miha-_> minus: sorry for noob questions, just for understanding. What does this mean in to buckets of 10000 numbers?
17:10:02 GMT <minus> http://redis.io/topics/memory-optimization this explains it i think
17:11:07 GMT <minus> basically you split your key into two parts instead of using it as a whole. the first part is the new key name, the second part is the field name in a redis hash.
17:14:34 GMT <miha-_> minus: but then for exp, I must think how to split numbers. I get that the main this is that if you have for exp two number: 1234567123 and 1234567567, i must split numbers like 123456 123 10, and 123456 567 10, right?
17:14:44 GMT <miha-_> the first purt must be the same
17:14:45 GMT <miha-_> ?
17:14:55 GMT <miha-_> minus: only this will optimize thing?
17:15:16 GMT <minus> are the numbers consecutive?
17:15:47 GMT <minus> what's the 10, your id you want to map to?
17:16:42 GMT <miha-_> minus: yes, but the issue is that this are all numbers for whole country :) it is hard to know in which block does it belongs to :)
17:17:22 GMT <miha-_> id is the number of operator this which number belongs to, this is for routing :)
17:17:39 GMT <minus> every time you need something for the phone number from redis you just split off the last 4 digits of the number to break it up, no magic
17:18:41 GMT <miha-_> minus: tnx
17:18:44 GMT <miha-_> i will try
17:18:48 GMT <miha-_> hope this will work
17:18:48 GMT <miha-_> :)
17:19:09 GMT <miha-_> minus: any other options to set on redis.conf?
17:19:21 GMT <minus> read the link
17:19:31 GMT <minus> it did mention something else
17:22:50 GMT <miha-_> minus: ok tnx :)
18:18:53 GMT <cloudbud> Can anybody tell me about redis clusters like a brief intro
19:40:02 GMT <BlackCrypt0> first time using redis on a cent 7 box.  I keep getting an error when trying to start redis that says Failed at step USER spawning /usr/bin/redis-server: No such process
19:43:05 GMT <minus> never seen an error like that; SELinux/AppArmor active?
19:49:17 GMT <BlackCrypt0> minus, this might be an SELinux issue, I disabled it and will see if that fixes it.
19:50:04 GMT <minus> i would've actually suggested to look at the logs or to set it to permissive mode only, but ok
20:04:44 GMT <BlackCrypt0> nope still getting the same error systemd[6145]: Failed at step USER spawning /usr/bin/redis-server: No such process
20:11:09 GMT <minus> maybe the user it's trying to start with doesn't exist?
20:11:27 GMT <minus> try starting redis-server from terminal to see if it even works?
20:19:26 GMT <BlackCrypt0> I figured it out...sudo instead of trying to start it as root...