In a system with no Internet access, following the instructions (to the letter) to upgrade from a 6.2.0 fabric to a 6.2.1 fabric takes a very long time – at least two hours.

The instructions are in s4.2 of this document:

https://access.redhat.com/documentation/en-US/Red_Hat_JBoss_Fuse/6.2.1/html/Migration_Guide/ESBMigrationPatching-Fabric.html

I started with a completely clean installation of Fuse 6.2.0, did fabric:create, and then create-ssh-container against a machine on the same subnet (a physical machine, not virtual). I then disabled all Internet access on both machines by editing /etc/resolv.conf to point to a non-existent DNS server.

The upgrade instructions work fine until it gets to running container-upgrade on the SSH container. The root container first shows the SSH container stuck in the "analysing" state for about ten minutes, and I see exceptions like this in the log of the SSH container:

Caused by: java.io.IOException: Error resolving artifact org.apache.cxf.karaf:apache-cxf:xml:features:3.0.4.redhat-621084: Could not transfer artifact org.apache.cxf.karaf:apache-cxf:xml:features:3.0.4.redhat-621084 from/to central (http://repo1.maven.org/maven2/): repo1.maven.org: Name or service not known
	at io.fabric8.maven.url.internal.AetherBasedResolver.resolve(AetherBasedResolver.java:512)[13:io.fabric8.fabric-maven:1.2.0.redhat-133]
	at io.fabric8.maven.url.internal.AetherBasedResolver.resolveFile(AetherBasedResolver.java:453)[13:io.fabric8.fabric-maven:1.2.0.redhat-133]
	at io.fabric8.maven.url.internal.AetherBasedResolver.resolveFile(AetherBasedResolver.java:438)[13:io.fabric8.fabric-maven:1.2.0.redhat-133]
	at io.fabric8.maven.url.internal.AetherBasedResolver.download(AetherBasedResolver.java:414)[13:io.fabric8.fabric-maven:1.2.0.redhat-133]
	at io.fabric8.maven.url.internal.Connection.getInputStream(Connection.java:124)[13:io.fabric8.fabric-maven:1.2.0.redhat-133]
	at org.apache.karaf.features.internal.RepositoryImpl.load(RepositoryImpl.java:97)[72:io.fabric8.fabric-features-service:1.2.0.redhat-133]
	... 18 more

The container does seem to be trying to contact repo1.maven.org. I also see error messages about repository.jboss.org:

| 13 - io.fabric8.fabric-maven - 1.2.0.redhat-133 | Retrying request to {s}->https://repository.jboss.org:443

However, I am not certain the exact point at which these messages are generated.

After a while, the SSH container's status changes to "downloading," where it gets stuck for a very long time. There are no log messages at all during this time, but a JVM stack dump of the SSH container reveals a Maven operation stuck in an HTTP read. So I guess it's another failed repository operation.