09:44:18 GMT <ningu> is there a reason redis would be dying with OOM when maxmemory is set, and maxmemory-policy is volatile-lru, and all keys have an expire set?
09:45:28 GMT <ningu> whoa weird. config says default is volatile-lru but somehow it's set to noeviction
19:33:42 GMT <ningu> isn't redis 3.2.6 out? the topic seems rather out of date
19:34:21 GMT <Habbie> ningu, it does look that way!
19:35:44 GMT <ningu> I am currently running 3.0.6 on Ubuntu 16.04. I don't use clustering and only use 2gb memory for a cache. all keys are strings (no lists, geo) with expires set. it seems like 3.2 doesn't have any particular features I need?
19:35:51 GMT <ningu> I don't know if performance would be much different
19:36:17 GMT <ningu> key values, I mean
19:36:32 GMT <minus> read the changelog and you will know
19:36:43 GMT <ningu> minus: I read the summary. the full changelog is quite long
19:36:48 GMT <Habbie> and performance does vary between versions
19:36:56 GMT <Habbie> i saw serious differences between 3.2.6 and git head (4-ish)
19:36:59 GMT <Habbie> not all of them positive
19:37:22 GMT <ningu> yeah... currently performance is fine, I suppose faster is always better
19:37:43 GMT <Habbie> my current pain is that redis is singlethreaded
19:37:48 GMT <Habbie> i sometimes just deploy 10 slaves on a machine
19:38:02 GMT <ningu> one per core?
19:38:23 GMT <minus> you were the powerdns guy, right?
19:38:34 GMT <Habbie> i am!
19:38:43 GMT <Habbie> ningu, something like that
19:39:15 GMT <*> minus  has more noise than actual queries
19:39:26 GMT <Habbie> noise?
19:39:28 GMT <Habbie> oh on DNS
19:39:36 GMT <minus> on redis
19:39:39 GMT <Habbie> oh
19:39:41 GMT <Habbie> why?
19:39:51 GMT <ningu> I could cluster if I really needed to
19:40:05 GMT <minus> well, not many requests
19:40:13 GMT <ningu> so say I have 10 cores and 2gb memory, does that mean 10 instances x 200mb each or something?
19:40:24 GMT <minus> pretty much, yes
19:40:30 GMT <Habbie> yes
19:40:34 GMT <ningu> ok
19:40:35 GMT <minus> that's a lot of cores for 2GB of RAM
19:40:39 GMT <Habbie> unless your OS does samepage mapping
19:40:41 GMT <Habbie> yes
19:41:13 GMT <Habbie> i'd love a redis feature where the parent process would fork off 10 new readonly children every 10-30 seconds
19:41:14 GMT <ningu> minus: it's 128gb total but a lot is dedicated to other things. it's a caching layer on top of a postgres instance. I realize we could spread this across other servers but it's a small operation we have going
19:41:18 GMT <Habbie> so i don't waste all that RAM
19:41:24 GMT <Habbie> i bet this would be useful to a lot of people
19:41:50 GMT <minus> i think so too
19:41:51 GMT <ningu> minus: we could spare more than 2gb honestly but so far it's been enough
19:41:56 GMT <minus> doesn't seem extremely hard either
19:42:01 GMT <Habbie> minus, right?
19:43:02 GMT <ningu> and there are 6 cores, not 12. I was just making up a number for the example
19:43:14 GMT <ningu> not 10*
20:30:48 GMT <dudeji> what should i do to increase performance of redis , i am getting alot of error saying "redis-server went away"