Hi Reza, Reza Housseini <[email protected]> skribis:
> I installing guix on our institutes cluster and followed the > instructions in this blog post [1]. But there are a couple of problems > I stumbled upon: > > - There is no mention how to mount the nfs shares on the nodes. I > added them to the /etc/fstab file but I'm unsure about the mount > options Just mount /gnu/store and /var/guix as NFS. The store should be mounted read-only; /var/guix must be writable. But that’s about it, no special options are needed. > - Thers is no guix binary on the nodes. I added a symlink to > /var/guix/profiles/per-user/root/current-guix/bin/guix in > /usr/local/bin Yes, that’s what needs to be done. > but when I try to install a package from a node I get an open-file > permission denied for > /var/guix/profiles/per-user/root/guix-profile.lock It looks like you’re running ‘guix’ on the node as root. Try running it as non-root instead. (If you really want to be able to run ‘guix’ as root on the nodes, then you need to pass the ‘norootsquash’ (not sure about the spelling) option for the /var/guix NFS mount. That’s not a great idea though.) > - In this email [2] it was mentioned I don't have to pass listen to > the guix-daemon, as the socket gets shared over nfs, but when I > remove the listen argument and try to install a package from a node > I get a failed to connect to /var/guix/daemon-socket/socket: > connection refused Unix-domain sockets cannot be shared over NFS. You have to use the ‘--listen’ option and set ‘GUIX_DAEMON_SOCKET’ as explained in <https://hpc.guix.info/blog/2017/11/installing-guix-on-a-cluster/>. HTH! Ludo’.
