Hello Openstack Discuss,
I am doing some investigation into instance entropy and was wondering what settings others are using with regard to rate limiting entropy supplied by the hypervisor.
Specifically, we're adding the "hw_rng:allowed=True" nova flavor property to pass libvirt the relevant config, but need to decide the appropriate rate limiting settings to prevent instances from being greedy with entropy but still retain a comfortable level for themselves. I've done some experimenting (100 bytes/s is possibly a minimum, but still allows a comfortable value of ~1000 for free entropy in instances)
I'm also curious to hear other's experiences when it comes to entropy in Openstack: * What sources of entropy did you use in the hypervisor? * Issues you've faced which was caused by insufficient entropy (instance or host)
Note, this is for a public cloud scenario, should that impact any suggestions you have.
Regards, Jordan