05:32:21 GMT <tzudot>  /part
10:32:18 GMT <antirez> Modules unloading get a bit tricky when they export data types...
11:47:21 GMT <manadart> Can I use CONFIG SET to change the bind value without restarting Redis?
11:48:08 GMT <manadart> A: No.
12:14:18 GMT <PSUser2> hi
12:14:42 GMT <PSUser2> is it possible to create a set of sets in redis?
12:15:18 GMT <PSUser2> for example set A: 1 2 3 set B: 2 3 4 set C: 2 1 3 should not be allowed because it's a duplicate of set A
12:19:28 GMT <Inge-> you can easily do set operations and determine that set A contains parts/all of set C
12:24:14 GMT <PSUser2> but if you have 2 million sets...
12:29:04 GMT <Inge-> It could take a little time :)
12:30:02 GMT <PSUser2> i wanted the built-inunique-ness of a hash
12:30:32 GMT <PSUser2> i guess that the only way would be to include the set, ordered from smallest to largest in the setname
12:30:56 GMT <PSUser2> this way, it wouldn't allow duplication
13:48:12 GMT <_Wise_> Hi *
13:48:16 GMT <_Wise_> up vote
13:48:16 GMT <_Wise_> 0
13:48:16 GMT <_Wise_> down vote
13:48:16 GMT <_Wise_> favorite
13:48:34 GMT <_Wise_> up vote
13:48:34 GMT <_Wise_> 0
13:48:34 GMT <_Wise_> down vote
13:48:34 GMT <_Wise_> favorite
13:48:53 GMT <_Wise_> (sorry about that)
13:50:33 GMT <_Wise_> I have several Redis 3.0 instances running in AOF persistence mode. Sometimes, I have to perform data migration from one instance to another.
13:50:36 GMT <_Wise_> Data migration process currently goes like this:
13:50:47 GMT <_Wise_> . BGSAVE on source instance and transfer rdb file to target instance
13:50:47 GMT <_Wise_> 2. restart target instance in RDB persistence mode
13:50:47 GMT <_Wise_> 3. wait for rdb file to be fully loaded
13:50:47 GMT <_Wise_> 4. trigger BGREWRITEAOF on target instance and wait for completion
13:50:48 GMT <_Wise_> 5. restart target instance in AOF persistence mode, and wait for database to be loaded
13:50:57 GMT <_Wise_> Is there a faster method ? Like a way to keep target instance in AOF persistence mode but instruct Redis to load an rdb file ?
13:51:06 GMT <minus> you could've just linked to the stack overflow post :)
13:51:16 GMT <_Wise_> true
14:00:21 GMT <SkiOne> hi, I am monitoring a situation. we are using redis with magento and over the last few days I have noticed a significant increase in key usage. I am trying to determine if its a real problem or not as the site is still behaving normally but when redis goes down Magento will crash
17:25:11 GMT <truthadjustr> hi, I'm playing on 3.2.0 sentinel as per documented where three boxes with setup [M1,S1], [R2,S2] and [R3,S3] and sentinel can do fail-over. But when I tried to completely separate put each into 6 boxes so that the sentinels are each separated, it's not failing-over eventually when i kill a master, revive it back and so on. I also see, thet revived master remains a master so now there are 2 masters in the cluster
17:29:40 GMT <truthadjustr> ok.. just found out that it failed-over eventually ..
17:29:44 GMT <truthadjustr> brb
17:37:29 GMT <truthadjustr> ok, so i guess what I was seeing was when I successively kill 2 masters one at a time, slowly..i was expecting that the remaining last 1 slave takes over but it didn't
18:10:09 GMT <jib> Hello!
18:10:58 GMT <jib> I've got an interesting problem I am trying to solve involving Redis. If anyone would be willing to lend some advice, I would be very grateful.
18:11:44 GMT <minus> hello jib. what exciting problems do you have to offer?
18:12:01 GMT <jib> We use Rackspace and last night their Dallas data-center had a switching loop and we lost connectivity to all of our cloud infrastructure. One of our Redis instances uses a mounted "Cloud Block Storage" volume which re-mounted as read-only. The only fix is to unmount, scan, and reboot, but there's new data in RAM that can't be saved to disk...
18:12:15 GMT <jib> Of course my first thought was to setup replication, but I am getting some strange errors.
18:13:23 GMT <jib> Primary, when I run a SYNC command, I get this: SYNC with master failed: -ERR Can't SYNC while not connected with my master
18:13:35 GMT <jib> Does replication rely on disk access to do its job or can I make it work from memory only?
18:15:48 GMT <minus> are you sure you got the connection settings right? (and the machine running the slave can reach it?) that'd be the obvious thing (from the message)
18:16:12 GMT <jib> Yeah, for sure. Both machines are on the same private network. I can ping and SSH both ways, and the ports are correct.
18:16:52 GMT <jib> I am seeing the following periodically in the logs: http://pastebin.com/Km6VDRQd
18:17:10 GMT <minus> can you redis-cli?
18:17:23 GMT <jib> Yeah, I have access to the CLI on both machines.
18:17:39 GMT <minus> oh
18:17:41 GMT <minus> i see
18:17:57 GMT <minus> i think it gathers data in a file before syncing
18:18:03 GMT <jib> Oh dear...
18:18:12 GMT <jib> Is there a way to force the dump/replication to go via RAM only?
18:18:52 GMT <jib> Perhaps even outside of Redis functionality, some way to even redirect where Redis wants to dump the RDB? Mounting an NFS perhaps?
18:18:55 GMT <minus> you can try enabling diskless replication (experimental)
18:19:05 GMT <minus> config set repl-diskless-sync yes
18:19:10 GMT <jib> Yeah, I saw that config and was just reading up on it. I'll take a closer look.
18:20:31 GMT <jib> Reading the docs, it makes sense why replication is not working. It's normally disk-backed which is currently read-only. I'll give the diskless option a try!
18:24:11 GMT <jib> Hmm. Same messages seen in the pastebin are spamming the logs on the slave.
18:24:28 GMT <jib> Could be that because it's a full sync and not a partial, it's too large for a diskless sync?
18:25:57 GMT <minus> sorry, no idea. i just knew the option exists
18:26:30 GMT <minus> you could try mounting something writable over /var/lib/redis
18:30:19 GMT <jib> Yeah, that's the next attempt. I already tried to attach a new cloud volume to the machine but because all of / is read-only, I'm having some trouble.
18:31:36 GMT <minus> you didn't run it in a master/slave setup?
18:32:26 GMT <jib> Not originally. I only just attempted to set that up right now, but there are issues with replication as we've seen.
18:35:05 GMT <jib> I think I got it working! I found a mounted drive which was still writeable.
18:35:35 GMT <jib> Thanks for your help, minus.
18:35:39 GMT <minus> what'd you do, a bind mount?
18:37:44 GMT <jib> I just noticed there was a tmpfs mounted with enough unused space.
18:38:05 GMT <minus> how did you get redis to save to there?
18:38:18 GMT <jib> In the CLI, I typed: config set dir ""
18:38:28 GMT <jib> And the next scheduled save started to dump to that new directory.
18:38:34 GMT <minus> oh, nice, didn't expect that to work during runtime
18:38:41 GMT <jib> Yeah, that's good to know!
18:38:56 GMT <minus> luckily you don't have SELinux or AppArmor on that machine
18:39:26 GMT <minus> anyway, good to see you got your data back
18:45:08 GMT <jib> Yeah, SELinux is set to permissive.
18:47:01 GMT <minus> any plans to turn it to enforcing?
18:51:26 GMT <jib> Not at the moment. The machine is not Web-accessible.