Home | Forums | Couchbase | Couchbase Server 1.8.x

port 8091 returned failure

4 replies [Last post]
Wed, 04/11/2012 - 05:14
yellowpaper
Offline
Joined: 04/10/2012
Groups: None

Hello,

My boss has set up 3 membase Server running v1.6. Now I am supposed to add a new Server to the cluster, preferable 1.8. So I installed an ubuntu server x86 on a virtual machine, added it to the network and then installed couchbase. I then used to the web-interface to add the server to the cluster but received the error "Attention - Got error response from remote node".
Afterwards I checked the error-log on the server that already is in the cluster and got following message:

During node join, the remote host "xxx.xxx.xxx.xxx" on port 8091 returned failure.

I don't really know how to solve it. Any help is appreciated.

Greetings,

yellowpaper

Top
Fri, 04/13/2012 - 00:46
ingenthr
Offline
Joined: 03/16/2010
Groups:

That would seem to indicate that something has gone wrong on the node you're trying to join.

Is there anything further in the log on the one you're trying to join to?

Top
Fri, 04/13/2012 - 03:33
yellowpaper
Offline
Joined: 04/10/2012
Groups: None

Nope, nothing in the logs. At least not in the webinterface under "Monitor"->"Log".

http://s14.directupload.net/images/120413/xesqehay.png

Top
Fri, 04/13/2012 - 09:51
ingenthr
Offline
Joined: 03/16/2010
Groups:

I actually had meant on the other side, where you're trying to join to.

That said, re-reading your original posting, you note that you're going from 1.6 to 1.8. You'll need to go to one of the 1.7 series first, 1.7.2 I think. Please have a look at the release notes for 1.8 and 1.7 at www.couchbase.com/docs

Top
Fri, 04/20/2012 - 04:09
yellowpaper
Offline
Joined: 04/10/2012
Groups: None

Thank you for your help so far. I will try to set it up the next few days. I just got one more question, because I got a certain method of updating in mind which I am not sure of if its recommended. What I want to do is not to loose any SESSIONS which are cached on our cluster + the cluster will move to 3 new machines. That's what I wanted to do:

We have 3x 1.6 Membase Servers.
I setup one 1.7 Server and add it to the cluster.
I then remove all 1.6 servers from the cluster step by step so it only consists of the one 1.7 server.
Afterwards I'd add 3x 1.8 Membase Servers to the cluster and remove the 1.7 again.

So it syncs data from 1.6 to 1.7 and then back to 1.8. Is that practicable? And would that ensure zero downtime on our membase cluster?

Greetings,

yellowpaper

EDIT.: Nvm, doesnt work. Just wanted to join 1.7.2 server to the 1.6 cluster. 1.6 Servers are on gentoo, 1.7.2 server is ubuntu. Might that be a problem?

EDIT2.: Log from my 1.7 Server which I tried to add t othe 1.6 cluster:

INFO REPORT  <5900.18472.5>                                 2012-04-19 16:24:42
===============================================================================
 
ns_log: logging menelaus_web:1:Invalid post received: {mochiweb_request,#Port<0.51761>,'POST',
                           "/addNodeRequest",
                           {1,1},
                           {6,
                            {"content-type",
                             {'Content-Type',
                                 "application/x-www-form-urlencoded"},
                             {"content-length",
                              {'Content-Length',"55"},
                              {"authorization",
                               {'Authorization',"Basic Og=="},
                               nil,
                               {"connection",
                                {'Connection',"keep-alive"},
                                nil,nil}},
                              nil},
                             {"te",
                              {"Te",[]},
                              {"host",{'Host',"213.183.3.249:8091"},nil,nil},
                              nil}}}}

I've tried to access IP:8091/addNoteRequest on all machines and it said "not found." Does this have something to do with it?

Top