I’ve tried now in both beta2 and beta3, I define a pool with a single master, or a pool with a master and 1 other slave and I end up with the same state. When the slave joins, it looks like the storage repos of the slave are known to the
master (local storage, removable storage, and dvd drives show up under the master not indented, but disconnected with red icon and slash), and the slave node is no where listed. When I connect to the console, it thinks it is connected. After a reboot, it doesn’t
know anything about its networking, storage repos, etc.
I’ve tried the emergency network reset and nothing has changed.
The slave believes it has a master, and any command line interactions with xe results in the message:
“The master says the host is not known to it. Perhaps the Host was deleted from the master's database? Perhaps the slave is pointing to the wrong master?”
… followed by the host UUID.
I’ve been able to reproduce this 100% of the time with both beta 2 and 3 of XCP 1.6. XCP 1.1 and 1.5 I can create a pool and join them together without issue.
The only thing that I do before joining these to a pool is removing the EXT3 ‘Local storage’ SR, and re-adding it back as an LVM SR. Is this another known bug?
- Brian Menges
Principal Engineer, DevOps
GoGrid | ServePath | ColoServe | UpStream Networks
