#rabbitmq IRC Log

Index

IRC Log for 2011-10-12

Timestamps are in GMT/BST.

[8:23] <KaZeR_W> hi there
[10:21] <Coffe> Hello. i am reading the guide on cluster. but there is noting about how then to query the cluster.
[10:35] <bob235> Coffe: do you mean finding out which nodes are in the cluster? see "rabbitmqctl cluster_status"
[10:42] <Coffe> bob2351, no. if i need to ad HA or common practis is to have a singel name to query , so it will be redundent if 1 node dies
[10:49] <bob235> Coffe: you can connect via a load-balancer, or you will need to inform your clients about all the nodes in a cluster
[10:50] <Coffe> bob235, ok. the load-balancer issingel point of failure .. tnx for the answers :)
[17:17] <ehm> I seem to have an issue where I delete a queue prior to the rabbitmq server going down and when it comes back up the queue is still there. I am using Pika and get the acknowledgement for the delete but maybe it is incorrect
[17:19] <ehm> rabbitmqctl says the queue has no consumers or messages too
[17:20] <ehm> p.s. by "acknowledgement" I mean I get a Queue.DeleteOk
[17:52] <bob235> ehm: does rabbit have anything interesting in the log when it starts up? how long do you wait between the delete-ok and shutting down the broker?
[18:01] <ehm> bob235: I did not check rabbit logs and have since overwritten them. Will try to trigger again. The shutdown was almost instantaneous and was a kill -9 on the server.
[18:11] <bob235> ehm: can you confirm the rabbit version?
[18:16] <ehm> bob235: version 2.3.1
[18:21] <antares_> bob235: hi. I have an implementation question, do you have a minute?
[19:17] <ehm> are Alternative Exchanges supported in version 2.3.1?
[19:20] <antares_> ehm: it is not clear from release notes for 2.3 and 2.2
[19:20] <antares_> you can try, it is a pretty easy to use extension
[19:25] <ehm> antares_: just tried and I got: 406:PRECONDITION_FAILED - inequivalent arg 'alternate-exchange' for exchange 'exchange_foo' in vhost '/': received none but current is the value 'my-ae' of type 'longstr'
[19:26] <ehm> I assume that means that alternate-exchange is not supported
[19:28] <antares_> ehm: no
[19:28] <antares_> ehm: it is supported
[19:28] <antares_> but you previously had it set to a different value
[19:28] <antares_> so rabbitmq thinks that arguments are different and rejects your declaration
[19:29] <antares_> delete that exchange using management UI or programmatically and try again
[19:29] <antares_> this happens to me every so often with 2.5 and 2.6 during development
[19:33] <old_sound> antares_: hey… have you seen this: https://github.com/videlalvaro/msg_store_bitcask_index
[19:35] <antares_> old_sound: I only have seen your announcement
[19:35] <antares_> old_sound: but this is just index, right?
[19:35] <old_sound> so far just index
[19:35] <antares_> not the complete store?
[19:35] <antares_> I need a redis-based store (not just index) but haven't gotten to it yet
[19:35] <antares_> there is a Riak one, too, to use as example
[19:35] <old_sound> where?
[19:36] <old_sound> why do you need a redis one?
[19:37] <antares_> to have queue introspection on the next travis-ci.org version
[19:39] <antares_> old_sound: the riak one is at https://github.com/jbrisbin/rabbit_riak_queue
[20:09] <old_sound> antares_: In pondering what's needed to be done to actually put riak_core in front of rabbit_msg_store
[20:09] <old_sound> I think the rabbit_msg_store is pretty god at storing messages
[20:10] <antares_> but why do you need riak_core there at all?
[20:10] <antares_> message store probably does not use consistent hashing (maybe in the case of clustering, I am not sure) or vector clocks
[20:14] <old_sound> I want the riak_core to do replication
[20:19] <antares_> so each node in a cluster will be responsible for a portion of the ring?
[20:20] <antares_> how do you hash messages in that case?
[20:22] <old_sound> I'm not 100% sure… I'm just reading about it… but I don't want to make the assumption of "hey let's put riak core" and scale this thing
[20:41] <lofidellity> Im attempting to get rabbitmq 2.3.1 in an HA active/active setup and I want to make sure I'm doing it correctly
[20:42] <lofidellity> right now i have two rabbitmq instances, running carp so they have one ip address in common and two seperate ip's as well of course
[20:42] <lofidellity> i have them setup so they are both disk nodes in a cluster
[20:43] <lofidellity> and when I create a queue I used this argument "x-ha-policy", "all"
[20:43] <lofidellity> is there anything else to do?
[20:45] <lofidellity> also one thing I've noticed in creating the cluster is that if I restart a node, it is unable to rejoin the cluster unless I restart the other node as well
[20:45] <lofidellity> how do I overcome this?
[22:23] <bob2351> lofidellity: HA active/active didn't exist in rabbit 2.3.1 - you will need a later version
[23:03] <lofidellity> bob2351: are there aditional configs available in later versions which make active/active possible because reading the docs make it look like active/active is really just having two brokers setup and marking a queue as mirrored
[23:04] <lofidellity> and for anyone curious as to why rabbit wouldnt start, I didn't have the other node in my /etc/hosts file so it couldnt connect
[23:16] <bob2351> lofidellity: HA active/active is more than a confi issue - the implementation is absent from the broker in releases prior to 2.6.0. see the changelog here: http://www.rabbitmq.com/changelog.html
[23:16] <lofidellity> awesome, I think this should lead me closer to what I'm looking for.
[23:16] <lofidellity> Thanks bob2351
[23:19] <lofidellity> are there any instructions availble for setup once I have the upgraded version?
[23:20] <lofidellity> So far all I was able to find was this: http://www.rabbitmq.com/ha.html
[23:52] <bob2351> lofidellity: that link should tell you everything and is the best source of information. as always ask here or on the mailing list if you get stuck

These logs were automatically created by MyxoRoboto on irc.freenode.net using the Java IRC LogBot.