[keycloak-user] [External] Re: 1.1 Beta2 in Wildfly cluster

[Schneider, John DODGE CONSULTING SERVICES, LLC]

I now have it working with my firewall enabled.  The Wildfly config is socket-binding with name “jgroups-udp”.  For an HA domain cluster, this is within socket-binding-group “ha-sockets”.  Default values are UDP port 55200 and multicast port 45688 with multicast address 230.0.0.4.  I think it would be helpful to mention this in the Keycloak docs.  The Wildfly docs for clustering only note information applicable to mod_cluster, which is different than this.

Thanks,
John

From: Schneider, John DODGE CONSULTING SERVICES, LLC
Sent: Friday, December 12, 2014 1:08 PM
To: 'Marek Posolda'; keycloak-user at lists.jboss.org
Subject: RE: [External] Re: [keycloak-user] 1.1 Beta2 in Wildfly cluster

Hi Marek,

Thanks for getting back to me.  I did see the ISPN000094 message you described in my log files, but it didn’t look like the messages you listed.  My messages only noted one node.  After disabling the firewall on both nodes, Keycloak is now working in domain mode with Infinispan providers in my config.  Now I just have to figure out all the ports necessary for JGroups to function correctly.  Once I figure this out, I will respond back.  Hopefully you can add this info to the documentation to help others out in the future.

Thanks again for your help,
John


From: Marek Posolda [mailto:mposolda at redhat.com]
Sent: Friday, December 12, 2014 6:56 AM
To: Schneider, John DODGE CONSULTING SERVICES, LLC; keycloak-user at lists.jboss.org<mailto:keycloak-user at lists.jboss.org>
Subject: [External] Re: [keycloak-user] 1.1 Beta2 in Wildfly cluster

Are you using shared database among both cluster nodes? Also when you start node1 and then start node2, you should see some message similar to this in the log of node1, which indicates that cluster nodes are connected:

wfnode_1 | 11:28:30,888 INFO  [org.infinispan.remoting.transport.jgroups.JGroupsTransport] (Incoming-1,shared=udp) ISPN000094: Received new cluster view: [wfnode1/web|1] (2) [wfnode1/web, wfnode2/web]
wfnode_1 | 11:28:33,767 INFO  [org.infinispan.remoting.transport.jgroups.JGroupsTransport] (Incoming-10,shared=udp) ISPN000094: Received new cluster view: [wfnode1/keycloak|1] (2) [wfnode1/keycloak, wfnode2/keycloak]



For more logging of which provider is used by keycloak-server.json, you can enable DEBUG logging for keycloak in standalone-full.xml (or domain.xml or whatever you are using):

                <logger category="org.keycloak">
                    <level name="DEBUG"/>
                </logger>

Also I think that editing file standalone/configuration/keycloak-server.json is just for standalone, but probably doesn't work for wildfly domain.


Maybe you can first try if cluster works in standalone configuration. If it helps, we can figure the domain later.

Marek

On 10.12.2014 00:57, Schneider, John DODGE CONSULTING SERVICES, LLC wrote:
Hi,

Correction, I *thought* everything was running in Wildfly domain mode.  It turns out I just got lucky by hitting the same server node in my initial test.  After a reboot and further testing today, I’m not able to login to the Keycloak admin console when both nodes in my cluster are running.  After attempting login, I am either taken back to a blank login page, or I see error “Unknown code, please login again through your application.”  Once in awhile, I can login without error. I should note that I’m using an Apache reverse proxy via mod_cluster.

I see no errors in the server logs.  I do see message “JBAS010281: Started <x> cache from keycloak container” for each of “realms”, “sessions”, “loginFailures”, “users”.  So, it looks like my domain config is working.  However, I can’t tell for sure that Keycloak is attempting to use the infinispan caches.  Some additional log output showing the values from keycloak-server.json would be helpful.  I used the CLI to upload “/profile=full-ha/subsystem=keycloak/auth-server=keycloak-1/:update-server-config(bytes-to-upload=/usr/local/wildfly/domain/configuration/keycloak-server.json~,overwrite=true)”  The response was “success” and then I restarted Wildfly on both nodes in the cluster.

Has anyone been able to get Keycloak 1.1 Beta 2 working in a wildfly domain, and using mod_cluster?  If so, could you please provide guidance?

Thanks,
John

From: Schneider, John DODGE CONSULTING SERVICES, LLC
Sent: Monday, December 08, 2014 6:43 PM
To: keycloak-user at lists.jboss.org<mailto:keycloak-user at lists.jboss.org>
Subject: 1.1 documentation update for running in domain HA mode

Hi guys,

Thanks so much for getting clustering support working in 1.1.  I have it up and running well in a Wildfly 8 domain setup under the “full-ha” profile.  One thing that I was pulling my hair out about for a while today were some errors related to Infinispan config.  I figured out that if running in HA cluster, you must include the “transport” element under the cache-container config (i.e. <transport lock-timeout=”60000” />).  It would be great if you could update Chapter 23 of the documentation to reflect this requirement.

Thanks,
John



_______________________________________________

keycloak-user mailing list

keycloak-user at lists.jboss.org<mailto:keycloak-user at lists.jboss.org>

https://lists.jboss.org/mailman/listinfo/keycloak-user

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/keycloak-user/attachments/20141212/0adb75f2/attachment-0001.html