node-v1.2.10

Runs against any current panel (still protocol version 4). Panel 1.2.16 and
later show the extra disks; an older panel ignores them and shows the system
disk as before.

Added

• Reports every local disk, not just /. The node reads
  /proc/self/mountinfo and lists the real disk filesystems — ext2/3/4, xfs,
  btrfs, zfs, f2fs, jfs, reiserfs, bcachefs, ntfs/ntfs3, exfat (and ntfs-3g on
  a block device), and vfat outside /boot — with their size and use, /
  first. tmpfs, overlay, squashfs, network and FUSE mounts, read-only loop
  images and anything under /boot are left out; a disk mounted twice (bind
  mounts, btrfs subvolumes, ZFS datasets of one pool) is listed once, under
  its shortest mount path; at most 16 are sent. The existing system-disk
  figures are unchanged. A disk that hangs on being read cannot stall the
  report or pile up threads: one collection runs at a time and reports carry
  the last finished list meanwhile. A node in Docker only sees the container's
  mounts.

Changed

• Installing and upgrading nodes on servers in mainland China. The node
  installer and the one-click self-upgrade now switch to accelerated downloads
  on their own when GitHub is slow or unreachable. Nothing to configure;
  servers that reach GitHub fine behave as before.