Cluster setup not working correctly with snarl (or maybe howl)

Description

Summary

I tried to setup a cluster of three fifo zones on three virtual machines on my local VMware environment. Both three fifo zones are installed based on the official documentation for version 0.8.2. All three zones contains all services howl, snarl, sniffle and ceberus.

After the installation was completed I connected all services to a cluster / ring. So if I followed the documentation correct all data (user information etc.) is shared between all three nodes.

I tried to login to the cerberus web interface an received random permission denied errors. After I disabled two snarl nodes everything is working correctly.

Expected behaviour

The ring should work for all services and the data should be shared in the cluster. It should be possible to login as user on cerberus without permission denied error if snarl is connected in the ring.

What I've done

Please have a look at the GitHub Gist.

Additional information

What really confusing me is the different output of howl-admin status and howl-admin member-status one command shows the information _ There are 3 handoffs pending!_ the other _ No pending changes._. Please have a look at the other GitHub Gist.

Environment

None

Status

Assignee

Heinz N. Gies

Reporter

Thomas

Labels

None

Components

Fix versions

Affects versions

Priority

Medium
Configure