01:39:08 GMT <Guest35026> Question, if I have an old redis running (version 2.6). But for some reason, whoever ran this, had placed the redis.conf file in /etc/, and the dir command has ./ set, and the instance is being run as non-root. So it is unable to create backups in /etc/ because it doesn't have permissions, and I can't even see what is going on because the logs are going to /dev/null, lol. I want to make a backup so I can fix all these things, but am
06:16:16 GMT <Inge-> Guest89879: sounds like fun. wonder if you could add a slave to it or if tyat also tries to write to /etc for the replication
07:24:06 GMT <twry> I'm getting an error when building redis from source: undefined reference to `clock_gettime' in function 'nstime_get' in /deps/jemalloc/lib/libjemalloc.a(nstime.o). any idea what I can do about this, please?
07:25:38 GMT <shmoon_> if i have 10GB memory and 100GB SSD/HDD - and I enable persistence in redis, will redis be able to store 100GB data ?
07:30:59 GMT <twry> never mind my question, had to patch the makefile and then it worked. thanks
08:29:11 GMT <badboy_> twry: yeah, known problem at the moment
08:29:35 GMT <badboy_> Inge-: it seems he is gone. next time: CONFIG SET dir /somewhere/redis/can/write/to
08:43:05 GMT <ren0v0> badboy_, when the master is back online, yes
08:43:20 GMT <ren0v0> you can set "priorities", so when all 3 are online it'll switch back
08:43:42 GMT <ren0v0> useful if you are using it for redundancy and have a more powerful box or something
08:45:07 GMT <ren0v0> badboy_, I'm considering reading from a local slave, instead of adding 300ms latency across the ocean to get to master. Is this possible with sentinel to find out the "local" slave?
08:55:10 GMT <badboy_> you can find slaves
08:55:23 GMT <badboy_> you have to decide client-side which one is local
09:04:52 GMT <Slydder> hey all
09:05:27 GMT <Slydder> is there a way to check if a DB is corrupt without having to do a dump?
10:00:21 GMT <Slydder> is there a way to check if a DB is corrupt without having to do a dump?
11:58:38 GMT <badboy_> Slydder: what exactly do you mean?
11:58:50 GMT <badboy_> corrupt in-memory data or a corrupt dump?
11:59:08 GMT <badboy_> because corrupt in-memory data means you got a whole lot other problems
11:59:27 GMT <badboy_> a corrupt dump however can come from a broken disk, broken transfer between servers or other means to corrupt persisted data
12:35:11 GMT <fxpester> what to do when redis-sentinel configured my master as slave it now prevent me from flushing it ? :D
12:35:45 GMT <fxpester> all my 3 nodes become slaves
12:38:12 GMT <badboy_> oO
12:38:23 GMT <badboy_> make one master manually and hope for the best?
12:38:32 GMT <badboy_> also what master are they replicating from?
12:41:17 GMT <fxpester> 1 master + 2 slaves, setuping my first redis cluster, so I did many restarts
12:50:05 GMT <badboy_> so what exactly is the problem?
12:51:02 GMT <Inge-> badboy_: Ah. good point.
16:39:39 GMT <bsamorim> what is redis' best practice when it comes to "connect many times" x "leave one single connection open the whole time" ?
17:08:59 GMT <badboy_> bsamorim: if you can re-use connections
17:09:14 GMT <bsamorim> badboy_:ok, thank you!
17:09:31 GMT <badboy_> tcp connection opening is kinda expensive if all you do is a single command
17:09:40 GMT <badboy_> so by re-using connections you avoid this cost
17:10:08 GMT <bsamorim> what if i'm connecting to localhost?
17:23:47 GMT <badboy_> localhost is cheaper, but still has overhead
17:24:15 GMT <badboy_> it still does some tcp-thingy
17:28:24 GMT <bsamorim> badboy_: k thx
18:21:16 GMT <Inge-> syn ack ack