Questions regarding Docker host path mounting.

[GoingTime]

When mounting a directory using Docker's host path mount feature—for instance, mounting the original path /mnt/foo/bar via the command /mnt/foo:/mnt/foo—I can only access the /mnt/foo directory itself but not the ./bar subdirectory, even though I have the necessary access permissions.

I have used various file management tools, yet this is the first time I have encountered such behavior. I would like to know whether this limitation stems from a specific security policy or if it is a bug.

[Hakanbaban53]

This is a well-known Docker behavior, and it's not a bug, it stems from how bind mounts interact with the host filesystem and Linux mount namespaces.

When you mount /mnt/foo:/mnt/foo, Docker creates a bind mount at that exact path inside the container. If /mnt/foo/bar is itself a separate mount point on the host (e.g., another disk, NFS share, loop device, or --bind mount), it won't be visible inside the container.

This is because Docker bind mounts are non-recursive by default. The Linux kernel's mount namespace only copies the top-level mount, not any sub-mounts nested beneath it.

[GoingTime]

I use ZFS, where each dataset acts somewhat like an independent mount point. Although I recursively passed through all the sub-paths using the --bind parameter—and can successfully see the datasets themselves—I still cannot see any files within those mount points.

I run nearly 20 containers and use host path mapping across all of them for data persistence and directory access—including tools with file management capabilities like Jellyfin, qBittorrent, and OpenList. I have never encountered this situation before, and it has me quite puzzled.