Troubleshooting Instance launch failure
Sometimes when you attempt to launch a NeCTAR instance, the Dashboard initially accepts the request only to have the launch fail and the instance go into Error state. This guide deal with a couple of common cases where there are possible workarounds.
"Error: Nova valid host was found"
This error typically means that the NeCTAR has attempted to launch the instance in an availability zone (AZ) that has insufficient capacity.
This can happen for a couple of reasons:
- You explicitly selected an AZ, and the AZ is full. (Or more precisely, it cannot identify a single compute node which has enough free cores, memory and local disk to satisfy the request.)
- You didn't select ana AZ, and there is something wrong with an AZ's NeCTAR node capacity information, and that has caused the "Nova" service to chose an AZ that is actually full.
The workaround is to try to get NeCTAR to launch in a different AZ:
- If you specified an AZ, try specifying a different one or "any availability zone". This may be inappropriate; e.g. if you need to launch in a specific AZ in order to access volume storage in that AZ.
- If you didn't specify an AZ, try specifying one.
Note that there is no guarantee that this will succeed. (A NeCTAR allocation does not guarantee that the capacity is available.)
"Error: Build of instance <id> aborted: Failure prepping block device"
This error can happen if you attempt to launch an instance using one of the "boot from ..." options that creates a new volume. If there is a delay in creating the volume, then the launch can fail. (This problem has been observed in OpenStack Juno.)
If the above scenario matches what you are doing, then a possible workaround is to create the new volume first via the Dashboard, wait for it to finish creating, and then do a simple "boot from volume".
If your launch attempts fail consistently with other errors, then you probably should create a NeCTAR Support request.