Hetzner Storage Box (CIFS)
-
I also switch all servers from CIFS to SSHFS and this reduced many problems.
+1 -
btw. If you have many SSHFS setups to do for many Cloudrons here is my script for creating some needed stuff.
This script creates a folder and ssh-key to be used with Cloudron.
This script expects to have the SSHFS to be mounted locally so it can create Folders and Keys which then can be used.#!/bin/bash read -rp "Customer Name: " CUSTOMER read -rp "Mount Point of SSHFS: " MOUNT_POINT CUSTOMER_PATH="${MOUNT_POINT%/}/$CUSTOMER" CUSTOMER_SSH_PATH="${CUSTOMER_PATH}/.ssh" CUSTOMER_AUTH_KEYS="${CUSTOMER_SSH_PATH}/authorized_keys" echo "=> Checking if customer already exists" if [[ -d $CUSTOMER_PATH ]]; then echo "=> Customer existing, quitting" exit 1 fi echo "=> Create Folder" mkdir -pv "$CUSTOMER_SSH_PATH" echo "=> Create Auth Keys File" if [[ ! -f "$CUSTOMER_AUTH_KEYS" ]]; then touch "$CUSTOMER_AUTH_KEYS" fi echo "=> Create ssh-key and place into auth keys" if [[ ! -f "${CUSTOMER_SSH_PATH}/$CUSTOMER" ]]; then ssh-keygen -f "${CUSTOMER_SSH_PATH}/$CUSTOMER" -t ed25519 -C "${CUSTOMER}-SSHFS" -N "" cat "${CUSTOMER_SSH_PATH}/${CUSTOMER}.pub" > "${CUSTOMER_SSH_PATH}/authorized_keys" else echo "=> Key seems to be present, something is off" exit 1 fi
-
From memory I think I once had a CIFS issue, but I think that was related to things changing on the Cloudron end (when Cloudron started automating mount stuff but I still had my slightly different set-up as per Hetzner docs), but I don't think I've ever had issues again since I set it up again with Cloudron automating the mount set-up.
Also, I spotted that apparently SSHFS is no longer maintained, so perhaps CIFS is again the best way/ only supported way?
@nebulon said in Best way/protocol to mount Hetzner Storage Box?:
https://github.com/libfuse/sshfs#this-project-is-orphaned unless the development was taken elsewhere without mention?
Also, I think the main problems lots of people were having previously (which effected CIFS/NFS/SSHFS) was fixed:
@girish said in Cloudron 7.3.4: "Analyze Disk" doesn't do anything and no statistics shown...:
Thanks all. This issue was already fixed as well last month - https://git.cloudron.io/cloudron/box/-/commit/67cde5a62cf0394c8bf2d78ee3408e5995a220e7 . It's already in 7.3.5.
Essentially, if you have CIFS/NFS/SSHFS, there is a crash.
-
@mdc773 said in Hetzner Storage Box (CIFS):
@robi has anyone figured out why CIFS is slow? why can't cloudron just allow volumes from objective storage
doesn't appear so.
Because object storage is not a filesystem and needs middleware to make it work. Things like rclone.org (which you can set up manually) or another gateway sw/hw that provides mounts for it.
-
Note that rclone mount is based on FUSE - https://rclone.org/commands/rclone_mount/ . So, symlinks etc won't work just like with s3fs.
-
-
-
I also had issues with CIFS in the past, switch to SSHFS and all has been well, except a couple of times on trying to restore a cloudron to another server. This just happened to me again yesterday where the restore the server from backup doesn't work (loading the backup config file) and I get the following error:
"failed to mount (inactive): read: Connection reset by peer"
The same error is also reported in this post.
For it to work I had to connect to CIFS.
Weirdly now, in the restore server, the same storage box connects fine as SSHFS for volumes (and I didn't do anything, Cloudron just restored my Volume config).
So personally, I think there is a bug somewhere in Cloudron that prevents mounting SSHFS for backup on Server restore, because mounting as SSHFS works fine otherwise.
-
Cloudron really just uses the native linux cifs tooling, but by experience cifs with hetzner every now and then results in a bad mountpoint, where a remount is required. So far the investigations why and when that happens were unfortunately inconclusive. At least for me the SSHFS option is a lot more stable.
-
@nebulon said in Hetzner Storage Box (CIFS):
At least for me the SSHFS option is a lot more stable.
Out of interest, have you testing doing a restore too as per @avatar1024 example above?
-
@nebulon This has been my exact experience too, except when trying to restore a server where SSHFS fails and CIFS works (which arguably is pretty important )
And of course here someone also had mounting issues with SSHFS.
SSHFS seems too good and reliable to be abandoned on Cloudron just yet, especially as while not actively maintained for new features / some issues, it is still maintained for critical issues.
-
I am also using SSHFS everywhere when I can