Re: [keycloak-user] SAML in a keycloak cluster

Hello Stian, Thank you for this hint and your help in general. Indeed you are right the URLs in the specified path returns security1 or security2 (Loadbalanced) and not security.lu. *<SingleSignOnService Binding="urn:oasis:names:tc:SAML:2.0:bindings:HTTP-POST" Location="http://mojito-security2.lu:8080/realms/master/protocol/saml <http://mojito-security2.lu:8080/realms/master/protocol/saml>"/>* *<SingleSignOnService Binding="urn:oasis:names:tc:SAML:2.0:bindings:HTTP-Redirect" Location="http://mojito-security2.lu:8080/realms/master/protocol/saml <http://mojito-security2.lu:8080/realms/master/protocol/saml>"/>* *<SingleSignOnService Binding="urn:oasis:names:tc:SAML:2.0:bindings:SOAP" Location="http://mojito-security2.lu:8080/realms/master/protocol/saml <http://mojito-security2.lu:8080/realms/master/protocol/saml>"/>* *<SingleLogoutService Binding="urn:oasis:names:tc:SAML:2.0:bindings:HTTP-POST" Location="http://mojito-security2.lu:8080/realms/master/protocol/saml <http://mojito-security2.lu:8080/realms/master/protocol/saml>"/>* *<SingleLogoutService Binding="urn:oasis:names:tc:SAML:2.0:bindings:HTTP-Redirect" Location="http://mojito-security2.lu:8080/realms/master/protocol/saml <http://mojito-security2.lu:8080/realms/master/protocol/saml>"/>* But our backend servers + the reverse proxy are configured according to specs + some rewrites. You can see our EAP and proxy configuration at the end of this messsage. The principal issue is that the backend servers are not able to understand themselves as *security.lu <http://security.lu>* but frankly I do not know how this could work as *we have nowhere configuration in the individual keykloack servers (EAP 7 + k* eycloak-overlay-1.9.8*) that reference the public name “security.lu <http://security.lu>”. *The *only machine* that knows the “security.lu” is the *RP *(apache + mod_proxy) which is hosted in another server. Could you explain how the individual Keycloak servers should automatically understand thay they are seen as “security.lu” from the outisde ??? What would be more logical is that the backend keycloack servers do have a configuration to know which other keycloack servers are on the same cluster and use this list as white-list to serve request regardless if the destination is security1 or security2 …. =============Standalone-ha XML config======= <subsystem xmlns="urn:jboss:domain:undertow:3.1" instance-id="mojitosecurity1"> <buffer-cache name="default"/> <server name="default-server"> <ajp-listener name="ajp" socket-binding="ajp"/> <http-listener name="default" proxy-address-forwarding="true" socket-binding="http" redirect-socket="proxy-https"/> <host name="default-host" default-web-module="keycloak-server.war" alias="localhost"> <filter-ref name="server-header"/> <filter-ref name="x-powered-by-header"/> <filter-ref name="proxy-peer"/> </host> </server> <servlet-container name="default"> <jsp-config/> <websockets/> </servlet-container> <handlers> <file name="welcome-content" path="${jboss.home.dir}/ welcome-content"/> </handlers> <filters> <response-header name="server-header" header-name="Server" header-value="JBoss-EAP/7"/> <response-header name="x-powered-by-header" header-name="X-Powered-By" header-value="Undertow/1"/> <filter name="proxy-peer" module="io.undertow.core" class-name="io.undertow.server.handlers.ProxyPeerAddressHandler"/> </filters> </subsystem> ….. <interfaces> <interface name="management"> <inet-address value="${jboss.bind.address. management:x.x.x.x.}"/> </interface> <interface name="public"> <inet-address value="${jboss.bind.address:192.168.xxx.xxx}"/> </interface> <interface name="private"> <inet-address value="${jboss.bind.address. private:192.168.xxx.xxx}"/> </interface> </interfaces> <socket-binding-group name="standard-sockets" default-interface="public" port-offset="${jboss.socket. binding.port-offset:0}"> <socket-binding name="management-http" interface="management" port="${jboss.management.http.port:xxxxx}"/> <socket-binding name="management-https" interface="management" port="${jboss.management.https.port:xxxx}"/> <socket-binding name="ajp" port="${jboss.ajp.port:xxxx}"/> <socket-binding name="http" port="${jboss.http.port:xxxx}"/> <socket-binding name="https" port="${jboss.https.port:xxxx}"/> <socket-binding name="jgroups-mping" interface="private" port="0" multicast-address="${jboss.default.multicast.address:xxx.x.x.x}" mu lticast-port="xxxxxx"/> <socket-binding name="jgroups-tcp" interface="private" port="xxxxx"/> <socket-binding name="jgroups-tcp-fd" interface="private" port=" xxxxx "/> <socket-binding name="jgroups-udp" interface="private" port=" xxxxx " multicast-address="${jboss.default.multicast.address: xxx.x.x.x }" multicast-port="45688"/> <socket-binding name="jgroups-udp-fd" interface="private" port=" xxxxx "/> <socket-binding name="modcluster" port="0" multicast-address=" xxx.x.x.xxx" multicast-port="xxxxx"/> <socket-binding name="txn-recovery-environment" port="xxxx"/> <socket-binding name="txn-status-manager" port="xxxx"/> <socket-binding name="proxy-https" port="xxx"/> <outbound-socket-binding name="mail-smtp"> <remote-destination host="localhost" port="xx"/> </outbound-socket-binding> </socket-binding-group> =============RP config=========== <VirtualHost *:443> ProxyRequests off ServerName security.lu SSLEngine On SSLProtocol ALL -SSLv2 -SSLv3 SSLCipherSuite xxxxxxxxxxxxxxxx; SSLHonorCipherOrder on <Proxy *> AddDefaultCharset Off Order deny,allow Allow from all </Proxy> <Proxy balancer://securitycluster> BalancerMember http://mojito-security1.lu:8080 route=mojitosecurity1 BalancerMember http://mojito-security2.lu:8080 route=mojitosecurity2 Order Deny,Allow Deny from none Allow from all # By default, ProxySet lbmethod=byrequests # ProxySet stickysession=ROUTEID </Proxy> <Location /balancer-manager> SetHandler balancer-manager # I recommend locking this one down to your # your office Order deny,allow Allow from 172.25.240.0/21 </Location> ProxyPass /balancer-manager ! ProxyPass / balancer://securitycluster/ stickysession=JSESSIONID| jsessionid RewriteEngine on <Location /realms/Tango/protocol/openid-connect> RewriteRule "/auth\?redirect_uri=https://facture\.lu/(.*)" "/realms/Tango/protocol/openid-connect/auth\?redirect_uri=http://billing2 \.lu/$1" Header edit Location http://billing2.lu https://facture.tango.lu </Location> <Location /realms/Tango/protocol/saml> Header edit Location ^http://mojito-security1.lu:8080 https://security.lu Header edit Location ^http://mojito-security2.lu:8080 https://security.lu <https://security.tango.lu> </Location> ErrorLog "/opt/csw/apache2/var/log/error_security.lu.log" TransferLog "/opt/csw/apache2/var/log/access_security.lu.log" </VirtualHost> Br Dimitrios Gkazgkas IT Solutions Architect ............................................................ .................................. *From:* Stian Thorgersen [mailto:sthorger@redhat.com] *Sent:* 20 October 2016 13:20 *To:* GKAZGKAS Dimitrios (TAN/MST) <Dimitrios.Gkazgkas(a)tangoservices.lu&gt; *Cc:* keycloak-user(a)lists.jboss.org; Benoît Reny <benoit.reny(a)its4u.lu&gt; *Subject:* Re: [keycloak-user] SAML in a keycloak cluster Check the urls in http://security.lu/auth/realms/master/protocol/saml/ descriptor. The URLs should contain security.lu and not URLs for the individual nodes. If that's not working, then you don't have the reverse proxy parts configured correctly. On 20 October 2016 at 11:47, GKAZGKAS Dimitrios (TAN/MST) < Dimitrios.Gkazgkas(a)tangoservices.lu&gt; wrote: Hello, This part of the configuration (Identifying Client IP Addresses" as well as "Enable HTTPS/SSL with a Reverse Proxy") is already in place in our system but still it does not work. Br Dimitrios Gkazgkas IT Solutions Architect ............................................................ .................................. *From:* Stian Thorgersen [mailto:sthorger@redhat.com] *Sent:* 19 October 2016 16:12 *To:* GKAZGKAS Dimitrios (TAN/MST) <Dimitrios.Gkazgkas(a)tangoservices.lu&gt; *Cc:* keycloak-user(a)lists.jboss.org; Benoît Reny <benoit.reny(a)its4u.lu&gt; *Subject:* Re: [keycloak-user] SAML in a keycloak cluster Hm.. Just reviewing that doc and it's not far from obvious. "Identifying Client IP Addresses" as well as "Enable HTTPS/SSL with a Reverse Proxy" are both relevant. On 19 October 2016 at 15:51, GKAZGKAS Dimitrios (TAN/MST) < Dimitrios.Gkazgkas(a)tangoservices.lu&gt; wrote: Hello, I suppose that you are talking about the part : Using the Built-In Load Balancer The thing is that if i understand well is that we can do this configuration for a domain clustered mode. Our configuration is currently a standalone clustered mode. This configuration can be also applied in this case ? Thanks for your reply, Br Dimitrios Gkazgkas IT Solutions Architect ............................................................ .................................. *From:* Stian Thorgersen [mailto:sthorger@redhat.com] *Sent:* 19 October 2016 14:36 *To:* GKAZGKAS Dimitrios (TAN/MST) <Dimitrios.Gkazgkas(a)tangoservices.lu&gt; *Cc:* keycloak-user(a)lists.jboss.org *Subject:* Re: [keycloak-user] SAML in a keycloak cluster If you configure your reverse proxy correct as well as configure it on the Keycloak side. Keycloak will see it's URL as security.lu and not the URL used by the reverse proxy to access it. The steps to do this is explained in the documentation I sent you. On 19 October 2016 at 14:29, GKAZGKAS Dimitrios (TAN/MST) < Dimitrios.Gkazgkas(a)tangoservices.lu&gt; wrote: ======Sent again without the picture===== Hello, Could you please be more specific ? In the documentation proposed it is referred how to FW the original client IP but our problem seems to be the Destination (IDP) inside the “samlp:AuthnRequest”. We get the following error: 2016-10-11 14:52:10,152 WARN [org.keycloak.events] (default task-2) type=LOGIN_ERROR, realmId=xxx, clientId=null, userId=null, ipAddress=xxxx, error=invalid_authn_request, reason=invalid_destination It seems to come from the following part of the code of Keycloack project. package org.keycloak.protocol.saml; public class SamlService extends AuthorizationEndpointBase *protected Response loginRequest(String relayState, AuthnRequestType requestAbstractType, ClientModel client) {* * SamlClient samlClient = new SamlClient(client);* * // validate destination* * if (requestAbstractType.getDestination() != null && !uriInfo.getAbsolutePath().equals(requestAbstractType.getDestination())) {* * event.detail(Details.REASON, "invalid_destination");* * event.error(Errors.INVALID_SAML_AUTHN_REQUEST);* * return ErrorPage.error(session, Messages.INVALID_REQUEST);* * }* The destination check simply do not much , request destination is always the internal keyclaock address “security1.lu” and it fails when saml requests end up to the second keycloack “securty2.lu”. Br Dimitrios Gkazgkas IT Solutions Architect ............................................................ .................................. *From:* Stian Thorgersen [mailto:sthorger@redhat.com <sthorger(a)redhat.com>] *Sent:* 18 October 2016 20:12 *To:* GKAZGKAS Dimitrios (TAN/MST) <Dimitrios.Gkazgkas(a)tangoservices.lu&gt; *Cc:* keycloak-user(a)lists.jboss.org *Subject:* Re: [keycloak-user] SAML in a keycloak cluster Please look at the documentation. It explains this. On 18 October 2016 at 16:57, GKAZGKAS Dimitrios (TAN/MST) < Dimitrios.Gkazgkas(a)tangoservices.lu&gt; wrote: Hello Stian, Thank you for your response. Could you explain a bit more what you mean by saying “*as Keycloak should see security.lu <http://security.lu>;, not the internal addresses of the nodes*” ? According to our understanding the Keycloak servers in the internal network is behind reverse proxy and thus they do not know that they are called “security.lu”, they just know that they are either security1.lu or security2.lu <http://security1.lu>;. When we tried to overwite the Saml XML configuration (that client uses for integration) and put the public address “security.lu” we again had the same ERROR in Keycloak logs “reason=invalid_destination” probably due to same root cause, the destination in the Saml AuthRequest was “Service.lu”, an address unknown for keycloack inside the private network. <samlp:AuthnRequest xmlns:samlp="urn:oasis:names:tc:SAML:2.0:protocol" Destination=" I attach our HA configuration. We do not use the build in Load Balancer but an Appache Reverse Proxy which actually rewrites all internall URLs to Publics for outgoing trafiif and the oposite for the incoming traffic. Thus there is not much left in the page you sent to be configured in our Keycloak. I hope I was clear. Any help would be highly appreciated. Br Dimitrios Gkazgkas IT Solutions Architect ............................................................ .................................. *From:* Stian Thorgersen [mailto:sthorger@redhat.com] *Sent:* 17 October 2016 20:41 *To:* GKAZGKAS Dimitrios (TAN/MST) <Dimitrios.Gkazgkas(a)tangoservices.lu&gt; *Cc:* keycloak-user(a)lists.jboss.org *Subject:* Re: [keycloak-user] SAML in a keycloak cluster Sounds like you haven't setup things properly as Keycloak should see security.lu, not the internal addresses of the nodes. Take a look at https://keycloak.gitbooks.io/server-installation-and- configuration/content/topics/clustering/load-balancer.html On 13 October 2016 at 19:14, GKAZGKAS Dimitrios (TAN/MST) < Dimitrios.Gkazgkas(a)tangoservices.lu&gt; wrote: The response from the list on my initial mails was : After content filtering, the message was empty So I try to send the same mail without CC and without attached =========== Hello, We are trying to configure a SAML authentication system in a keycloak cluster. First, with only one node , we are currently managing to authenticate in SAML way. The architecture : --> we have one apache reverse proxy with a public and unique endpoint for saml authentication. We can call the pubic url : security.lu< http://security.lu> --> the reverse proxy will load-balance all calls that come on security.lu <http://security.lu> to two keycloak nodes : security1.lu< http://security1.lu> and security2.lu<http://security2.lu> ( the private urls) . The issue that we have : --> The client that integrates saml has a tomcat and integrates a keycloak-saml.xml file. Of course, in this file the configuration is refering to security1.lu<http://security1.lu> ( the private address as the keycloak node only knows its private address). --> If we arrive during the load-balancing on the security1.lu< http://security1.lu> node, it will work. If I arrive on the second security2.lu<http://security2.lu> node, it will fail. When I dig a little bit more, it's because in fact, the SAMLRequest that is generated looks like this : <samlp:AuthnRequest xmlns:samlp="urn:oasis:names:tc:SAML:2.0:protocol" Destination="http://security1.lu<http://security1.lu>:8080/realms/xxx/protocol/saml" ForceAuthn="false" ID="ID_e563f50b-4ed8-454c-b938-0727d18ec08e" IsPassive="false" IssueInstant="2016-10-11T12:52:09.865Z" Version="2.0"><saml:Issuer xmlns:saml="urn:oasis:names: tc:SAML:2.0:assertion">xxxxx</saml:Issuer><samlp:NameIDPolicy AllowCreate="true" Format="urn:oasis:names:tc:SAML:2.0:nameid-format: persistent"></samlp:NameIDPolicy></samlp:AuthnRequest> The error that I get is an invalid_destination because we receive this SAMLRequest on the security2.lu<http://security2.lu> node : 2016-10-11 14:52:10,152 WARN [org.keycloak.events] (default task-2) type=LOGIN_ERROR, realmId=xxx, clientId=null, userId=null, ipAddress=xxxx, error=invalid_authn_request, reason=invalid_destination >From what I see there is for saml client, a Clustering tab where I have currently nothing. Maybe I need to add some host nodes here ? But i don't know how to proceed. Or is there any way to define both security1.lu<http://security1.lu> and security2.lu on the Saml XML configuration that the client integrates? We have set proxy-address-forwarding=true Thank you for your help. Kr, Br Dimitrios Gkazgkas IT Solutions Architect ________________________________ **** DISCLAIMER **** http://www.tango.lu/maildisclaimer _______________________________________________ keycloak-user mailing list keycloak-user(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/keycloak-user