02:43:24 GMT <gitgud> hi is there an official site for redis benchmarking?
02:43:35 GMT <gitgud> i want to know how many read/second and write/second can be done?
07:49:36 GMT <minus> gitgud: redis-benchmark is shipped with redis, if that helps
07:50:17 GMT <gitgud> ok thanks minus
08:09:49 GMT <svandyk> gday - newbie to redis and busy experimenting on EC2/Azure. I'm still reading through the docs, so apologies if this is covered in there.
08:10:15 GMT <svandyk> how do we tell an app that connects to redis to connect to a slave in case of master going down?
09:28:38 GMT <arnaud_> morning
09:36:08 GMT <arnaud_> I have a strange (well at least for me) behaviour with my redis cluster... I got some ZADD triggering an OOM message:  OOM command not allowed when used memory > 'maxmemory'
09:36:31 GMT <arnaud_> but when I look at all the nodes memory information, the peak memory was never greater than the maxmemory
15:11:13 GMT <moldy> hi
15:13:11 GMT <moldy> on a new machine, redis doesn't start. https://dpaste.de/3nm4
15:13:20 GMT <moldy> any idea what would typically cause this?
15:13:40 GMT <moldy> i am running the same provisioning code against other servers where it works fine
15:15:36 GMT <moldy> oh i should add that this is on boot
15:17:06 GMT <moldy> ah i think i got it. never mind... not enough coffee.
15:25:09 GMT <minus> My guess is that you bound redis to an interface that was not up yet
15:41:06 GMT <moldy> minus: the problem seems to be that /var/run/redis must exist
15:41:19 GMT <moldy> i am not sure who should create it though -- the debian package?
15:41:36 GMT <moldy> hmmm, it's a tmpfs
15:42:08 GMT <moldy> so it seems that on some servers, /var/run/redis is created at the right point in time, and on this one it is not
15:42:10 GMT <minus> debian 8+? then with systemd's tmpfiles
15:42:24 GMT <moldy> both the working and the non-working systems are debian8
16:00:45 GMT <moldy> minus: ah actually it fails on the other systems too :)
16:05:24 GMT <minus> deterministic behaviour, always nice to have
16:06:27 GMT <moldy> i'm tempted to just work around it by placing the pid file in /tmp
16:11:47 GMT <minus> set up tmpfiles.d properly
16:12:00 GMT <minus> it's just a couple of lines
16:13:23 GMT <moldy> i'm using somebody else's ansible role that seems to download, compile and redis itself :p
16:13:31 GMT <moldy> *and install
16:22:14 GMT <minus> eh
16:22:26 GMT <minus> otoh, debian. outdated versions are no fun
16:22:35 GMT <minus> otooh debian 8 ships with 3.0
16:22:40 GMT <minus> iirc
16:22:44 GMT <minus> using that too
17:18:51 GMT <X-warrior> When using LUA with redis, it does receive KEYS and ARGV, what is the special case with KEYS? I can only change data refereed on keys? so If I'm creating a lua command which will change a queue and a sset I need to pass both names?
17:19:57 GMT <X-warrior> a list and a sorted set*
17:21:57 GMT <badboy_> X-warrior: in normal Redis there is no technical difference
17:22:10 GMT <badboy_> in Cluster mode it is essential as only KEYS is checked for being in the same slots
17:23:02 GMT <X-warrior> ok, so it is safer to pass them directly instead of calculating them inside my lua function to guarantee I will be able to use cluster mode in future
17:23:03 GMT <X-warrior> thanks
17:31:19 GMT <badboy_> well, if you never target cluster it doesn't matter
17:31:30 GMT <badboy_> and there are valid use cases for calculated-on-the-fly keys
17:32:19 GMT <minus> they should be valid as long as the created keys are not meant to be accessed from clients, right?
17:32:54 GMT <badboy_> uhm, why not?
17:34:23 GMT <minus> In Cluster Mode
17:35:26 GMT <minus> Well, never mind
17:37:06 GMT <badboy_> well, in Cluster mode you _have_ to use KEYS
18:18:36 GMT <X-warrior> badboy_, minus in my use case I'm thinking to create something that replicates a data structure I need, a list for a queue, a sorted set to find expired tokens easily on cleanup and a hash to find it easily on direct access...
18:19:30 GMT <X-warrior> so I was thinking to have some lua functions which keep this structure syncrnized and use just a name and concatenate it as my_struct:__NAME__:list, my_struct:__NAME__:sset, etc..
18:34:17 GMT <X-warrior> brb