Systemd takes the advantage of autofs to do the 'mount-on-demand', thus leaves to entries in the mount table for one filesystem (one for the real filesystem, one for the autofs). And this behavior results in mlocate skipping the filesystem since autofs is in PRUNEFS (and I actually don't feel it's wrong for autofs to be there).
attachment fix-multiple-mount-entries.patch
Thanks for the patch.
I suppose, to be precise, updatedb should base the decision on the "innermost" mount (the one that is actually visible). This information is available only in /proc/self/mountinfo , so fixing this will depend on some other changes (necessary to fix bind mount handling).
Metadata Update from @mitr: - Issue assigned to mitr
Metadata Update from @mitr: - Assignee reset
Login to comment on this ticket.