When I kill a restic mount with Ctrl-C something is going wrong causing the repo mount point path to become unusable showing “Transport endpoint is not connected” when I try re-mounting the same directory or even executing ls. Further, ls shows lots of ??? for that mount point.
The mount point I’m aiming at is located inside an NFS hosted home directory. My current workaround is to login from a different VM which is also mounting that NFS home directory and then rmdir that mount point. Then I can make a new mount point directory. The rub which makes this very challenging is that I can’t make that new mount point directory with the same name. If I keep the same name, I again see the ??? and get the same FUSE error. I would like to script the mounting process so that I can restart restic mount at the same mount point if it dies.
I haven’t seen anyone having this issue with restic, but I’ve been searching google for a while trying to find similar problems with other FUSE tools. I tried using fusermount -uz $FOLDER but that never works because it does seem like mount point is being unmounted, just not very cleanly! The same goes for using umount directly. Does anyone have any suggestions for how to deal with the issue described above? Are there workarounds I can do so that I don’t have to rename the mount point? I really want to have a standard script which I can apply to many different restic repos all mounted on the NFS share simultaneously in the same way.
I’m using the stock restic from the CentOS7 repository. I found this blog post suggesting that FUSE mounting on NFS folders just isn’t currently well supported. I am hoping I am reading that wrong - since what restic gives me with FUSE mounts is really valuable to me and works better and easier than I could have imagined once it’s running.