Project

General

Profile

Incorrect ClusterRepoItem, skipping connection attempt...

Hamid Alimohammadi
Added over 3 years ago

Hello,

I had tested a two node Tigase clustering some months ago and it was working,

but when I try it now on new servers I get an error in tigase-console "WARNING: Incorrect ClusterRepoItem, skipping connection attempt.."

To find the reason I installed two fresh Tigase using console installer with minimal configuration but the same error.

Do you know what may cause the problem?

The followings are the 'init.properties file', 'WARNINGS in tigase-console.log' , 'and database output for cluster_nodes'

node1:

--cluster-mode = true
config-type = --gen-config-all
--debug = server,db,cluster
--user-db = mysql
--admins = admin@node1.domain.com
--user-db-uri = jdbc:mysql://node2.domain.com/tigasedb12?user=tigase&password=tigase12
--virt-hosts = xmpp.domain.com
--sm-plugins = -message-archive-xep-0136,+jabber:iq:auth,+urn:ietf:params:xml:ns:xmpp-sasl,+urn:ietf:params:xml:ns:xmpp-bind,+urn:ietf:params:xml:ns:xmpp-session,+jabber:iq:register,+jabber:iq:roster,+presence,-jabber:iq:privacy,-jabber:iq:version,+http://jabber.org/protocol/stats,+starttls,-msgoffline,-vcard-temp,-http://jabber.org/protocol/commands,-jabber:iq:private,-urn:xmpp:ping,-basic-filter,-domain-filter,-pep,-zlib

node2:

--cluster-mode = true
config-type = --gen-config-all
--debug = server,db,cluster
--user-db = mysql
--admins = admin@node2.domain.com
--user-db-uri = jdbc:mysql://node2.domain.com/tigasedb12?user=tigase&password=tigase12
--virt-hosts = xmpp.domain.com
--sm-plugins = -message-archive-xep-0136,+jabber:iq:auth,+urn:ietf:params:xml:ns:xmpp-sasl,+urn:ietf:params:xml:ns:xmpp-bind,+urn:ietf:params:xml:ns:xmpp-session,+jabber:iq:register,+jabber:iq:roster,+presence,-jabber:iq:privacy,-jabber:iq:version,+http://jabber.org/protocol/stats,+starttls,-msgoffline,-vcard-temp,-http://jabber.org/protocol/commands,-jabber:iq:private,-urn:xmpp:ping,-basic-filter,-domain-filter,-pep,-zlib

WARNING in node1:

2015-05-12 15:00:42.317 [main]             ClusterConnectionManager.itemAdded()  WARNING: Incorrect ClusterRepoItem, skipping connection attempt: node1.domain.com:5ab3779020749007d6e2514eef5e26b828c73450ea6df6740965d9e8e19670a7:5277:0:0.0:0.0

WARNING in node2:

2015-05-12 14:49:17.869 [main]             ClusterConnectionManager.itemAdded()  WARNING: Incorrect ClusterRepoItem, skipping connection attempt: node2.domain.com:86f5910a8bc32f608119af48464aeda15da903dfea2bd6dc6aa77dbbb7c2b529:5277:0:0.0:0.0

Database output for cluster_nodes:

+-------------+------------------------------------------------------------------+---------------------+------+-----------+-----------+
| hostname    | password                                                         | last_update         | port | cpu_usage | mem_usage |
+-------------+------------------------------------------------------------------+---------------------+------+-----------+-----------+
| node1.domain.com | 5ab3779020749007d6e2514eef5e26b828c73450ea6df6740965d9e8e19670a7 | 2015-05-12 15:01:01 | 5277 | 0.3974563 | 24.897293 |
| node2.domain.com | 86f5910a8bc32f608119af48464aeda15da903dfea2bd6dc6aa77dbbb7c2b529 | 2015-05-12 15:01:01 | 5277 | 0.3117531 | 22.029015 |
+-------------+------------------------------------------------------------------+---------------------+------+-----------+-----------+

I read in a recent document that there is no need to add --cluster-nodes so I omitted it.

The DNS names are resolved properly and round robin DNS records has been created for node1.domain.com , node2.domain.com and virutal host xmpp.domain.com.

Just to add the 'password' filed in DB for cluster_nodes does not change when I refresh the query while to my understanding it should be changed(?).

Thank you in advance for help


Replies (3)

Added by Wojciech Kapcia TigaseTeam over 3 years ago

The warnings may be a bit misleading - they are only informative and correct - ClusterRepoItem is ignored on the same machine in order to avoid connecting nodes to themselves (which would break clustering) - I've changed a bit log information to avoid confusion.

As for the clustering problem - yes, nodes information should be updated periodically in the database (vide last_update column).

Please check if Tigase opens cluster connection (either netstat or lsof will do). If yes, please verify that they are correctly established (i.e. handshaking is correct, in the logs).

Added by Hamid Alimohammadi over 3 years ago

Thank you for reply,

The last_update column is updated periodically and properly in database, but the password fields are always fixed strings, in my case:

Password for node1.domain.com is always '5ab3779020749007d6e2514eef5e26b828c73450ea6df6740965d9e8e19670a7'

Password for node2.domain.com is always '86f5910a8bc32f608119af48464aeda15da903dfea2bd6dc6aa77dbbb7c2b529'

Is it ok and normal that nodes passwords are always fixed in database?

If the reply is yes, then my minimal clustering is working fine and sorry for my mistake saying "I got an error..."

Added by Wojciech Kapcia TigaseTeam over 3 years ago

Yes, it's ok passwords are fixed as they are generated only once during startup of the instance.

    (1-3/3)