fabric8io/fabric8-devops

message-broker App on fabric8 connection to jolokia failed

shoaibjdev opened this issue · 10 comments

shoaibjdev commented

After creating the App for message-borker, activemq and zookeeper, I tried to connect to the AMQ console once the pods were up, but it seems its refusing to connect to the JVM with below error:

Error: 'EOF'
Trying to reach: 'http://172.17.0.6:8778/jolokia/?canonicalNaming=false&ignoreErrors=true&maxCollectionSize=500&maxDepth=7'

Do note, i'm able to connect to other App consoles, it's just AMQ console which is throwing error.

rawlingsj commented

Hmm I'm not too sure about this, I wonder if the message broker is using an old base image + old jolokia - @rhuss do you recognise this issue? Do you think it's base image related?

shoaibjdev commented

In the recent build for gofabric8 which is 0.4.121, none of the Apps seems to work. I have tried messaging-gateway and apiman both fails to connect to JVM via jolokia connector.

I have tried on different platforms (Mac minikube, CentOS 2 nodes minion cluster, Fedora 2 nodes) still same result.

I have also tried using earlier versions like gofabric8-linux-amd64–0.4.112, gofabric8-linux-amd64–0.4.20 still same issue with Jolokia connector.

Can someone please advise on this as starting development on fabric8 seems challenging now.

rhuss commented

@rawlingsj As the console tries to reach the Jolokia agent here via the pod (?) ip from the browser, this obviously doesn't work. The connection should go over the API server proxy like for OpenShift, but I'm afraid this feature was never really tackled

shoaibjdev commented

@rhuss: fabric8-forge seems to be connecting and working via jolokia connector which is installed by default as part of fabric8 installation. If it's using pod ip to reach Jolokia agent this should also not connect. Also if I create an application such as spring boot camel app based on CI/CD flow, i'm able to connect to its JVM using same jolokia connector. Please advise.

rhuss commented

Sorry I have no fabric8 installation right now to test or to debug (and unfortunately not much resources left to help much). What are the steps I would need to reproduce ?

shoaibjdev commented

@rhuss: Thank you very much for responding.
Below are the steps to reproduce.

  1. Download latest gofabric8 release (0.4.121) from https://github.com/fabric8io/gofabric8/releases
  2. chmod +x gofabric8
  3. gofabric8 start
  4. Once the fabric8 is up and console is accessible
  5. Default Team dashboard -> Runtime -> Run (Icon on the top right)
  6. Type messaging to filter out message queue related apps
  7. Select messaging and Run on the top right
  8. After few minutes when the images are pulled (messaging-gateway, mqbroker and zookeeper) and pods are running
  9. Try connecting to the JVM console for any of messaging application like say mqbroker (There is small icon on the pod list under Runtime for connecting to JVM console)
rhuss commented

I just sent a PR as fabric8io/fabric8-ipaas#581 to disable the openshift authentication in which case we should be good again. For testing fabric8/message-broker:2.2.168-1 and fabric8/message-gateway:2.2.168-1 has been pushed to docker hub.

rhuss commented

@rawlingsj feel free to review and trigger a release if happy.

ChadCorsentino commented

has this issue been resolved yet? I'm having the exact same issue in minikube. I have version 2.4.24 of the fabric8-platform installed.

pbuditi commented

I have same problem as well. has this been fixed already ?