HOAB

Problem :

It seems that I have no metric on one particular method while it works for all the others.

This is method is  :

public final boolean myMethod(myArgs) 

Does the fact that this method is final is a problem for bytecode instrumentation of Introscope ?

Solution :

No. It works :) As usual. My problem is somewhere else.

3 classes : 

Parent

package com.test.caapm.finalmethodtest;

public class ParentClass {
    public void finalMethod() {
        System.out.println("parentFinalMethod");
    }
}

Middle

package com.test.caapm.finalmethodtest;

public class TestFinalMethodAgent extends ParentClass {

    public static void staticMethod() {
        System.out.println("staticMethod");
    }

    public final static void finalStaticMethod() {
        System.out.println("finalStaticMethod");
    }

    public final void finalMethod() {
        System.out.println("finalMethod");
    }

    public static void main(String... args) {
        TestFinalMethodAgent test = new TestFinalMethodAgent();

        while(true) {
            test.finalMethod();
            TestFinalMethodAgent.staticMethod();
            TestFinalMethodAgent.finalStaticMethod();
            try {
                Thread.sleep(1000);
            } catch (InterruptedException e) {
                throw new RuntimeException(e);
            }
        }
    }

}

Child

package com.test.caapm.finalmethodtest;

public class ChildClass extends TestFinalMethodAgent{
}

Pbd

SetFlag: TestFinalMethod
TurnOn: TestFinalMethod

IdentifyDeepInheritedAs: com.test.caapm.finalmethodtest.ParentClass TestFinalMethod

TraceAllMethodsIfFlagged: TestFinalMethod PerIntervalCounter "{classname} - {method}"

Problem :

When using Spring Oauth2 resource server, it checks the received token, to do so it must retrieve token validaty or get certificates to validate token. These two calls are usually made in HTTPS for obvious security reason.

When the certificate is not known by the Spring Oauth2 resource server JVM, it fails as SSL handshake can not complete. It fails. Without a single message :) 

Solution :

Put the following log level on : 

org.springframework.security.oauth2.server.resource.web.BearerTokenAuthenticationFilter: DEBUG

I see that the exception is propagate to authenticationFailureHandler I see that it should be handle by authenticationFailureHandler but somewhere in the chain, it's not traced… Could search deeper next time.

It will activate this log : 

            try {
                AuthenticationManager authenticationManager = this.authenticationManagerResolver.resolve(request);
                Authentication authenticationResult = authenticationManager.authenticate(authenticationRequest);
                ...
            } catch (AuthenticationException var11) {
                SecurityContextHolder.clearContext();
                if (debug) {// 134
                    this.logger.debug("Authentication request for failed!", var11);
                }
                this.authenticationFailureHandler.onAuthenticationFailure(request, response, var11);
            }

Problem :

I use Keycloak 19.0.1 behind a proxy (nginx) and wasn't able to connect to the admin part of keycloak.

With a reverse proxy nginx and keycloak, login in admin console lead to be blocked on :

/realms/master/protocol/openid-connect/login-status-iframe.html/init?client_id=security-admin-console ....

With a 204 return code and no other errors. 

Solution :

I had to explore keycloak source code to find the cause ; This test failed in keycloak.js : if ((event.origin !== loginIframe.iframeOrigin) in keycloak.js

After a (lot of) time of search, it appears that it compares : https://mydomain/keycloak and https://mydomain:443/keycloak because I'd setup hostname-port to 443 in keycloak.config.

My keycloak configuration : 

hostname=mydomain  
proxy=reencrypt  
hostname-strict=false  
hostname-port=443  
hostname-path=keycloak  
http-relative-path=keycloak  
hostname-admin-url=https://mydomain/keycloak

So keycloak build his URL as follow : https://mydomain:443/

And the browser send : https://mydomain/ as 443 is a default port and not displayed in the URL.

By removing the port, it works perfectly : 

#hostname-port=443

I open a discussion to improve documentation here

Problem :

I tried to connect Broadcom Introscope 10.7 and SAML given by Keycloak.

Based on these documents : 

• Forum
• Official guide

Well not enough to make it works.

Solution :

Thanks to remote debug mode, the key is that the callback URL is : 

https://<webview url>/saml.jsp

Search for saml.jsp + introscope on google. Good luck.

Here are the steps (assuming that you already have a keycloak realm up and ready) : 

Step 1 : IntroscopeEnteprise.properties

introscope.saml.enable=true
introscope.saml.request.binding=POST
introscope.saml.idpUrl=<URL_KEYCLOAK>/realms/<your realm>/protocol/saml
introscope.saml.issuer=com.ca.apm.webview.serviceprovider
introscope.saml.webstart.issuer=com.ca.apm.webstart.serviceprovider
introscope.saml.em.issuer=com.ca.apm.em.serviceprovider
introscope.saml.principalAttributeName=principalName
introscope.saml.groupsAttributeName=groups
introscope.saml.webstart.tokenTimeoutInSeconds=60
introscope.saml.internalIdp.enable=false
# introscope.saml.internalIdpUrl=http://localhost:8080/idp/profile/SAML2/POST/SSO

Step 2 : Keycloak configuration

1. Create a client named as introscope.saml.issuer so in our case :  com.ca.apm.webstart.serviceprovider
2. Enter the callback URL in Master SAML Processing URL : https://<webview url>/saml.jsp

Step 3 : Certificates

You should secure you communication between Introscope and Keycloak : 

1. Provide HTTPS for Keycloak
2. Provide HTTPS for Introscope
3. Sign information in Keycloak client
4. Import Keycloak key in a JKS truststore for Java (Webview part). Keycloak client certificate are in the client definition, tab “Keys”.
   1. Follow Official guide to create the JKS
5. Point to this truststore (example : spprivatekey.jks) - next steps.

Step 4 : IntroscopeWebview.properties:

apm.webview.saml.sp.truststore=/path/to/spprivatekey.jks

Problem :

Launching a JVM with Introscope agent on a old JVM 1.7 result in : 

A problem occurred while attempting to create the delegate agent 
[IntroscopeAgent] Agent Unavailable 

Well, as the agent is heavily customized, removing customization was the first step. It starts. 
OK, a few lambda removing alter, I recompiled everything with Java 1.7 target.. And still the same message. 

Solution : 

After a hours ... 

Step 1 : 

• Decompile
  • com.wily.introscope.agent.AgentShim
• Add more logs, it confirms this is a class loading problem with major/minor version. I finally get the class name 

Step 2 : 

• After checking X times my maven settings, I finally used : *
  • javap -v | grep version
  • to check if it was ok or not. 
• and surprisingly, the generated class was OK ! 
• but I used also assembly plugin 

Step 3 : 

• The cause was that I recompile some classes of an old jar and rebuild it 
• I tracked the faulty classes version in the different repositories  The root cause was that I installed in a local repository the old jar without the modified classes with the command : 

plaintext mvn install:install-file -DcreateChecksum=true -Dfile=./agent/wily/Agent.jar -DgroupId=com.ca.agent -DartifactId=javaagent -Dversion=10.0.7SP3 -Dpackaging=jar -DlocalRepositoryPath=local-maven-repo

which install it also after build in the global repository... 

And assembly plugin use in priority the global repository. 

Well I didn't take time to understand why maven do not use only my local repository for this jar. 

I just add : rm -rf ~/.m2/repository/com/ca/agent/sqlagent in the beginning of my install script  ! 

Well a few hours lost again here...