fabric:unjoin?

[Javier]

I've been searching for an answer on the internet but have yet to find it.  Once doing a fabric:join, if I do a shutdown -f and then run fusefabric.bat again, I can't rejoin the fabric.

I don't see a way a command to unjoin a fabric before shutting down.  So how do I "unjoin" the fabric prior to shutting down fusefabric?

-Javier

[Ioannis Canellos]

I've been searching for an answer on the internet but have yet to find it.  Once doing a fabric:join, if I do a shutdown -f and then run fusefabric.bat again, I can't rejoin the fabric.

This is not the expected behavior. It sounds like a bug to me. Could you please provide some more details regarding your setup?

 

There is no unjoin command, but you can reuse join to join the same or another fabric whenever you want.

--
Ioannis Canellos

FuseSource

Blog: http://iocanel.blogspot.com

Twitter: iocanel

[Javier]

Running fuse-fabric-7.0.1.fuse-083

Machine A

Windows 7 Enterprise SP1, 64 Bit OS, 32 bit Java 6 JDK, 8 GB RAM,  8 Cores

Machine B

Windows 7 Enterprise SP1, 64 Bit OS, 32 bit Java 6 JDK, 6 GB RAM,  4 Cores

On Machine A:

Run fusefabric.bat, run fabric:create, create a profile using a feature we've developed in-house, and create a child container running on Machine A using the newly created profile.  Everything works as expected.

On Machine B:

Run fusefabric.bat, run fabric:join MachineA:2182.  Using the profile defined on Machine A as the profile type, create a child container on Machine B.  The container on Machine B works as expected.  (Which machine I use to crate the child container doesn't seem to matter.)  Issue command shutdown -f.  Re-run fusefabric.bat and try to run fabric:join MachineA:2182 and I get the following error: "A container with the name: MachineB is already member of the cluster. You can specify a different name as an argument."

Restarting fusefabric.bat on Machine A doesn't make the problem go away.  To rejoin, I've wiped fusefabrc from Machine A, re-installed it, re-run it, and re-created my profile on Machine A.  At that point, I can rejoin Machine B.

[Ioannis Canellos]

On Tue, Aug 21, 201On Machine B:

Run fusefabric.bat, run fabric:join MachineA:2182.  Using the profile defined on Machine A as the profile type, create a child container on Machine B.  The container on Machine B works as expected.  (Which machine I use to crate the child container doesn't seem to matter.)  Issue command shutdown -f.  Re-run fusefabric.bat and try to run fabric:join MachineA:2182 and I get the following error: "A container with the name: MachineB is already member of the cluster. You can specify a different name as an argument."

There is also -f option which will force the same container name.

However, you shouldn't need to do so. The root container on machine B should automatically rejoin fabric. If you can share some more details, like full container log or anything that would help on identifying why the rejoin doesn't happen automatically, I could raise an issue about it.

[Javier]

Do you want the karaf.log from Machine A, MachineB?  Or the log files from the instances under each one?

[Ioannis Canellos]

The full log of machine B would be enough I think.