On Fri, Jun 14, 2019 at 12:33:43PM +0200, Wouter Verhelst wrote: > On Thu, Jun 13, 2019 at 10:55:36AM -0400, Josef Bacik wrote: > > Also I mean that there are a bunch of different nbd servers out there. We have > > our own here at Facebook, qemu has one, IIRC there's a ceph one. > > I can't claim to know about the Facebook one of course, but the qemu one > uses the same handshake protocol as anyone else. The ceph ones that I've > seen do too (but there are various implementations of that, so...). > Ah, for some reason I remembered Qemu's being distinctly different. I suppose if most of the main ones people use are using the same handshake protocol that makes it more compelling. But there'd have to be a really good reason why a initramfs isn't viable, and so far I haven't heard a solid reason that's not an option other than "it's hard and we don't want to do it." > > They all have their own connection protocols. The beauty of NBD is > > that it doesn't have to know about that part, it just does the block > > device part, and I'd really rather leave it that way. Thanks, > > Sure. > > OTOH, there is definitely also a benefit to using the same handshake > protocol everywhere, for interoperability reasons. > Sure, Facebook's isn't different because we hate the protocol, we just use Thrift for all of our services, and thus it makes sense for us to use thrift for the client connection stuff to make it easy on all the apps that use disagg. Thanks, Josef
