The usual phrase “Cannot dispatch VM to any Host. Possible reasons: Not enough capacity in Host or System DS, dispatch limit reached, or limit of free leases reached.” is not informative at all. It always requires us to log in to the server and start investigating scheduler’s logs to figure out why it can’t instantiate a VM.
Could we have the scheduler log related to that VM printed out in a tab in sunstone? That way we could scroll through and see which reason prevented the VM from launching on a specific server.
Not sure if this is possible. As you said in other thread, you have more then 50 hosts. One may be discarded due to memory, second not enough CPU, third wrong hypervisor, …
The error message can be quite big, if we want to include all information.
I’ll look to the code to check if the logging can be improved.
Yes the log would be big, I can imagine that. Especially with debug level 5 So this requirement would require planning on what exactly would be printed, and whether or not it would contain all the retries. If the VM is in PENDING state for 24 hours, it would accumulate a lot of logs from every cycle it tried to deploy.
Created an issue for this: Better Scheduler messages in case VM not dispatched · Issue #5489 · OpenNebula/one · GitHub
I found a way how to report at least this reasons: No host, No network, No datastore