Julian Hjortshoj

That all makes sense. Thanks very much for the quick and thoughtful feedback. We will have a think about how we want to proceed.

Interesting....so, how does the server know what "this client" means? In our use case, we may have numerous fuse-nfs processes starting and stopping on the same host, potentially mounting the...

That's excellent to hear! NFSv4 and file locking have both been on our roadmap for a while. If we can add support for v4, then I think support for locking...

FWIW, id mapping and multithread don't appear to change the behavior. I also get the same behavior without those parameters: ``` root@8cb85777-ed10-4e6a-5e5f-651c612dade9:/tmp/build/5bbc57d5# fuse-nfs -a -n "nfs://testnfsserver.gorgophone.cf-app.com/export2/certs?version=4" -m /tmp/volumes/nfs-volume-name root@8cb85777-ed10-4e6a-5e5f-651c612dade9:/tmp/build/5bbc57d5# touch...

the nfs3 version still works fine: ``` root@8cb85777-ed10-4e6a-5e5f-651c612dade9:/tmp/build/5bbc57d5# fuse-nfs -a -n "nfs://testnfsserver.gorgophone.cf-app.com/export2/certs" -m /tmp/volumes/nfs-volume-name root@8cb85777-ed10-4e6a-5e5f-651c612dade9:/tmp/build/5bbc57d5# touch /tmp/volumes/nfs-volume-name/grak4 root@8cb85777-ed10-4e6a-5e5f-651c612dade9:/tmp/build/5bbc57d5# echo "food" > /tmp/volumes/nfs-volume-name/grak4 root@8cb85777-ed10-4e6a-5e5f-651c612dade9:/tmp/build/5bbc57d5# echo "food" >> /tmp/volumes/nfs-volume-name/grak4 root@8cb85777-ed10-4e6a-5e5f-651c612dade9:/tmp/build/5bbc57d5# echo "food"...

so....if we remove the version file, where do we maintain a record of the current version? I had the notion that folks might consume VERSION in their build scripting, given...

Deprecating the capability until the next major release would allow us to avoid making a breaking change, I think. Plugins using the value could continue to advertise the capability, and...

@resouer I believe that Saad was referring to #96

I feel like we discussed this possibility early on, and eventually discarded it because some plugins will require fuse software or other non-kernel mounting protocols that we don't necessarily want...

Fair enough. So if I understand it correctly, basically what you're saying is that the container runtime would always need to do a bind mount to consume the volume as...