Description of problem: Networking between build pods and imagestream pod are not the same. This causes issues with when OpenShift is behind a proxy as shown below: I0910 10:39:55.756609 1 sti.go:393] Downloading: https://repo1.maven.org/maven2/org/jboss/bom/jboss-javaee-6.0-with-tools/1.0.0.M1/jboss-javaee-6.0-with-tools-1.0.0.M1.pom E0910 10:39:58.807202 1 sti.go:419] Sep 10, 2015 10:39:58 AM org.apache.http.impl.client.DefaultHttpClient tryConnect E0910 10:39:58.807785 1 sti.go:419] INFO: I/O exception (java.net.NoRouteToHostException) caught when connecting to the target host: No route to host Version-Release number of selected component (if applicable): 3.0.1 How reproducible: Very Steps to Reproduce: 1. Establish a SQUID and block traffic except trough the proxy 2. install Basic 1 Broker, 2 Node OSE setup. 3. Establish a build: using https://github.com/bit4man/kitchensink-html5-mobile-example (or similar kitchen sink build) - Note: Changes to the following were made to enable proxy http://post-office.corp.redhat.com/archives/openshift-sme/2015-August/msg01004.html Actual results: Builds Fails: I0910 10:39:55.756609 1 sti.go:393] Downloading: https://repo1.maven.org/maven2/org/jboss/bom/jboss-javaee-6.0-with-tools/1.0.0.M1/jboss-javaee-6.0-with-tools-1.0.0.M1.pom E0910 10:39:58.807202 1 sti.go:419] Sep 10, 2015 10:39:58 AM org.apache.http.impl.client.DefaultHttpClient tryConnect E0910 10:39:58.807785 1 sti.go:419] INFO: I/O exception (java.net.NoRouteToHostException) caught when connecting to the target host: No route to host Expected results: STI is to complete the build of https://github.com/openshift/kitchensink-example - Note: Changes to the following were made to enable proxy http://post-office.corp.redhat.com/archives/openshift-sme/2015-August/msg01004.html Additional info: # docker ps CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 496ddc7de255 registry.access.redhat.com/jboss-eap-6/eap-openshift:latest "/bin/sh -c 'tar -C 7 minutes ago Up 7 minutes 8080/tcp, 8443/tcp elated_hopper 0c47e5aed093 openshift3/ose-sti-builder:v3.0.1.0 "/usr/bin/openshift- 7 minutes ago Up 7 minutes k8s_sti-build.1b8182a7_kitchensink-6-build_demo_44093bf0-588a-11e5-bdcd-525400efa961_57803a89 60f113a6172a openshift3/ose-pod:v3.0.1.0 "/pod" 7 minutes ago Up 7 minutes k8s_POD.892ec37e_kitchensink-6-build_demo_44093bf0-588a-11e5-bdcd-525400efa961_8da4fc48 Doing docker exec on the LAST of the two build containers fails to use the proxy. The docker exec on the FIRST of the build container succeeds. [jboss@496ddc7de255 ~]$ curl -x http://192.168.100.5:3128 http://cnn.com curl: (7) Failed connect to 192.168.100.5:3128; No route to host [jboss@496ddc7de255 ~]$ exit [root@node2 ~]# docker exec -it 0c47e5aed093 /bin/bash [root@kitchensink-6-build openshift]# curl -x http://192.168.100.5:3128 http://cnn.com <html> <head><title>301 Moved Permanently</title></head> <body bgcolor="white"> <center><h1>301 Moved Permanently</h1></center> <hr><center>nginx</center> </body> </html>
Sending to networking since the pod can't reach the proxy.
Can you include level 5 build logs so we can confirm the http proxy is being set? but i don't think that's the issue since your docker exec shows the proxy is fundamentally unreachable from the build pod.
When I was on the call we definitely saw maven was trying to access the proxy. It said something like "Fetching <some url> via <the proxy>:3128"
This seems to be an issue with moving systems and altering IP addresses.