Differences

This shows you the differences between two versions of the page.

--- services:fileserver [2024/09/05 09:23] – [Fileserver access via WebDAV] willy
+++ services:fileserver [Unknown date] (current) – removed - external edit (Unknown date) 127.0.0.1
@@ Line 1: / Line 1: @@
-====== File Server ======
-I will not discuss how to share your files on the home network using __legacy__ tools like [[https://it.wikipedia.org/wiki/Network_File_System|NFS]] or [[https://it.wikipedia.org/wiki/Samba_(software)|SAMBA]], there are plenty of tutorials online and, beside, it's kind out of the scope for self-hosting.
-I will focus on how to provide access via __web browser__ and via __WebDAV__, which is a web-based sharing protocol a bit like NFS or SAMBA, but aimed ad broader __internet__ access, and not __intranet__ access.
-The idea is to create share areas where your users will be able to store files. It is possible to extend this idea also to user-specific areas where each user can put private stuff not visible by other users, but this require a little bit extra complexity and might be addressed in the future.
-You will be using your SSO authentication, there will be no need to create new users anywhere, and it will of course be protected by the Reverse Proxy for external access.
-A future upgrade might prefer the use of //sFtpGo// instead of //FileBrowser + Apache//.
-===== Overall Architecture and Shares =====
-This solution leverages two tools:
-  * [[services:filebrowser|FileBrowser]], to provide browser access to your shares
-  * [[https://httpd.apache.org/download.cgi|Apache]] strictly as WebDAV server
-The NGINX reverse proxy will integrate with your preferred [[selfhost:sso|SSO]] authentication and add the HTTPS layer to ensure all access is properly encrypted.
-I will assume that your shares are under **/shares**, but of course each share can be located anywhere you like. Let's also assume, as an example, that your share is called __/shares/common__ and is managed by the user __fileserver__ of the group __users__. The requirement for users and groups will be detailed later on.
-Each share folder will have the following structure:
-  * /share/common/: main folder
-  * /share/common/db: contains the FileBrowser database, needed for browser access
-  * /share/common/webdav: contains WebDAV specific stuff, for WebDAV access
-  * /share/common/data: contains the actual shared files and directories.
-This structure is provided as an example to follow, of course you can move the individual folders where you prefer. The only caveat is that, for security reasons, the **db** and **webdav** folder should **not** be inside the **data** folder.
-You will also need to assign two ports for each share, as an example for our //common// share:
-  * 3002: port for FileBrowser
-  * 10001: port for Apache WebDAV server
-Any other share can start from these port numbers and go up in numbering.
-I choose to assign a dedicated subdomain, **drive.mydomain.com**, as file server and organize the shares like this:
-  * **https://drive.mydomain.com/common**: direct browser access URL for __common__
-  * **https://drive.mydomain.com/webdav/common**: WebDAV specific access URL for __common__
-  * **https://drive.mydomain.com**: will show a main directory page to access the shares.
-You can add any more folders as separate shares as you like. Due to how WebDAV works, it is mandatory to separate the browser accessible URLs from the WebDAV ones, like i did above.
-=== Permissions and Users ===
-Each share will be accessible by different users, so this needs to be planned a bit. For user-specific shares, not much needs to be done except run FileBrowser for the specific share as the specific user. This is left as an exercise for you.
-For common shares instead, it's important to create one common user, which i will call **fileserver** user to run the associated services, and create the **/home/common** folder.
-You need to assign that folder to the **users** group and the **fileserver** user:
-<code bash>
-useradd -d /shares/common -m fileserver -g users
-</code>
-You need to set the //umask// for the fileserver user to **0002** so that any new files created by it will be writable by the users. Also, create the **db** folder, where the FileBrowser database will need to be located, and the **webdav** folder, there specific Apache configuration need to be located, and of course don't forget the **data** folder, where you can put the shared content itself:
-<code bash>
-su - fileserver
-echo "umask 0002" >> ~/.bashrc
-source ~/.bashrc
-mkdir db
-mkdir webdav
-mkdir data
-</code>
-===== Fileserver access via Browser =====
-I am currently using [[services:filebrowser|FileBrowser]] because it's lightweight, don't get in the way, is flexible and simple to use. Check the linked page for the generic installation instructions, and here are the specific details for this case.
-I assume FileBrowser has been installed on your system already by following my guide.
-You will need to run **one** instance of FileBrowser for //each share//, so you will to allocate one specific port for each share. I will describe how to run it for the **common** share, so //FileBrowser// will run as the **fileserver** user that you created above.
-So, create the specific **/etc/conf.d/fileserver.common**:
-<file - filebrowser.common>
-BASE_URL="/common"
-DATABASE="/shares/common/db/filebrowser_common.db"
-DESCRIPTION="Common web archive"
-FOLDER="/shares/common/data"
-GROUP="users"
-PORT=3002
-USER="filebrowser"
-</file>
-Create the **init.d** symlink too, and start it. Of course, choose a free port (3002). See my [[services:filebrowser|FileBrowser]] instructions page.
-===== Fileserver access via WebDAV =====
-__NOTE:__ using HTTP will cause a 301 redirect to HTTPS, and WebDAV clients will fail. So use HTTPS URL in webdav clients and not HTTP.
-While there are a few WebDAV servers like [[https://github.com/micromata/dave|Dave]], they seems to be either unmaintained or overly complicated. Also NGINX can be a WebDAV server, but it seems to be buggy and not supporting LOCK stuff, so i decided to go with Apache web server, which also has a long standing WebDAV implementation.
-The idea here is to run a dedicated copy of Apache as user //fileserver// and group //users// so that it can access and manage the shared files. So first you need to emerge apache:
-<code bash>
-emerge apache
-</code>
-WebDAV is enabled by default in Gentoo Apache ebuild, so there is no need to fix USE flags.
-You will **not** be running Apache as system service, because that will mess with our user permission approach. I have prepared the following init script that manages to start separated Apache copies for each of your shares. Do drop the following file to **/etc/init.d/webdav**:
-<file - webdav>
-WD_LOG_PATH="/var/log/webdav"
-WD_SLOT="${SVCNAME#webdav.}"
-WD_USER=${USER:-${WD_SLOT}}
-WD_GROUP=${GROUP:-${WD_SLOT}}
-WD_SHARE=$(basename ${WD_FOLDER:-xxx})
-WD_ROOT="${WD_FOLDER}/webdav/root"
-WD_MOUNT="${WD_ROOT}/webdav/${WD_SHARE}"
-description=${DESCRIPTION:-WebDAV starter}
-pidfile="/run/${RC_SVCNAME}.pid"
-apache_args=(
--c "ServerRoot /usr/lib64/apache2"
--c "LoadModule dav_module modules/mod_dav.so"
--c "LoadModule dav_fs_module modules/mod_dav_fs.so"
--c "LoadModule dav_lock_module modules/mod_dav_lock.so"
--c "Include /etc/apache2/modules.d/*.conf"
--c "User ${WD_USER}"
--c "Group ${WD_GROUP}"
--c "DavLockDB ${WD_FOLDER}/webdav/locks"
--c "PidFile ${pidfile}"
--c "ErrorLog ${WD_LOG_PATH}/${WD_SLOT}/error.log"
--c "CustomLog ${WD_LOG_PATH}/${WD_SLOT}/access.log common"
--c "DocumentRoot ${WD_ROOT}"
--c "ServerName 127.0.0.1"
--c "Listen 127.0.0.1:${PORT}"
--c "<Directory ${WD_ROOT}>"
--c " DAV On"
--c " AllowOverride All"
--c " Options -Indexes +FollowSymlinks -ExecCGI -Includes"
--c " Require all granted"
--c "</Directory>"
--c "SetEnv redirect-carefully"
-)
-start_pre() {
-        if [ -n ${WD_FOLDER} ]
-        then
-                if [ -d ${WD_FOLDER} ]
-                then
-                        if [ "${WD_SLOT}" != "webdav" ]
-                        then
-                                ebegin "Checking log path..."
-                                test -e "${WD_LOG_PATH}" || mkdir "${WD_LOG_PATH}"
-                                test -e "${WD_LOG_PATH}/${WD_SLOT}" || {
-                                        ebegin "Creating log path '${WD_LOG_PATH}/${WD_SLOT}'"
-                                        mkdir "${WD_LOG_PATH}/${WD_SLOT}"
-                                } && chown -R ${WD_USER} "${WD_LOG_PATH}/${WD_SLOT}"
-                                ebegin "Checking root path '${WD_MOUNT}'"
-                                test -e ${WD_MOUNT} || {
-                                        ebegin "Creating root path '${WD_MOUNT}'"
-                                        mkdir -p ${WD_MOUNT}
-                                        chown ${WD_USER}:${WD_GROUP} ${WD_MOUNT}
-                                }
-                                ebegin "Checking root mount/bind '${WD_MOUNT}'"
-                                test -z "$(mount | grep ${WD_MOUNT})" && {
-                                        ebegin "Mounting/binding root path '${WD_FOLDER}/data' -> '${WD_MOUNT}'"
-                                        mount -o bind ${WD_FOLDER}/data ${WD_MOUNT}
-                                }
-                                eend 0
-                        else
-                                ebegin "Error: do not run this script, run a link to it!"
-                                eend 255
-                        fi
-                else
-                        ebegin "Error: '${WD_FOLDER}' does not exist!"
-                        eend 255
-                fi
-        else
-                ebegin "Error: FOLDER must be defined!"
-                eend 255
-        fi
-}
-start() {
-        start-stop-daemon -w ${WD_TIMEOUT} --start --pidfile "${pidfile}" -- \
-                /usr/bin/apache2 "${apache_args[@]}"
-        eend $?
-}
-stop_post() {
-        ebegin "Checking root mount/bind '${WD_MOUNT}'"
-        test -n "$(mount | grep ${WD_MOUNT})" && {
-                ebegin "Unmounting/unbinding root path '${WD_FOLDER}/data' -|-> '${WD_MOUNT}'"
-                umount ${WD_MOUNT}
-        }
-        eend 0
-}
-</file>
-and make it executable:
-<code bash>
-chmod +x /etc/init.d/webdav
-</code>
-=== Create apache configuration files for each share ===
-By using the above init script, defining a new share means to create a share symlink of that script and the associated config file.
-For our __common__ example share, create the following **/etc/conf.d/webdav.common**:
-<file - webdav.common>
-DESCRIPTION="Common WebDAV archive"
-FOLDER="/shares/common"
-GROUP="users"
-PORT=10001
-USER="filebrowser"
-</file>
-Note the port, it needs to be unique and available.
-Create the symlink:
-<code bash>
-ln -s /etc/init.d/webdav /etc/init.d/webdav.common
-</code>
-=== Prepare apache folders for each share ===
-The above mentioned init script will create all the needed sub-folders for you, but here is a recap:
-  * /share/common/.apache/root: which will map as WebDAV root (you will see why)
-  * /share/common/.apache/locks: which will be used for WebDAV lock databases
-You can manually create them if you like:
-<code bash>
-su - fileserver
-mkdir /home/common/.apache/root
-mkdir /home/common/.apache/locks
-</code>
-=== Messing with the WebDAV root folder ===
-Now, the fun part is that you want to protect this behind the NGINX reverse proxy (for HTTPS and authorization reasons) and it seems that WebDAV does **not** play well with URL redirection and similar funny things. In other words, the base url you will be using on the reverse proxy **must match** the url in the Apache. You **cannot use** rewrite directives or Alias stuff.
-Since you will be exposing the browser-based access as **https://drive.mydomain.com/common** and the WebDAV access as **https://drive.mydomain.com/webdav/common** it means that we need to connect your **/shares/common/data folder to /shares/common/webdav/root/webdav/common** for it to work. Nicely messed up eh?
-Since symbolic links cannot be used by WebDAV (could it be //that// simple?), the only viable option is **mount -o bind**. This is taken care automatically in the above init script.
-=== Startup Apache for the share (and autostart) ===
-Since you have already created the share specific startup script symlink and the associated config file, all you need to do is add it to the default runlevel and start it:
-<code bash>
-rc-update add webdav.common default
-/etc/init.d/webdav.common start
-</code>
-===== Reverse Proxy and wrap-up =====
-Everything is protected behind the [[selfhost:nginx|NGINX Reverse Proxy]], so you should create the following config file for the **drive** subdomain:
-<file - drive.conf>
-server {
-        server_name drive.mydomain.com;
-        listen 443 ssl;
-        listen 8443 ssl;
-        http2 on;
-        access_log /var/log/nginx/drive.mydomain.com_access_log main;
-        error_log /var/log/nginx/drive.mydomain.com_error_log info;
-        # WebDAV requires basic auth, while normal auth can be used with FileBrowser
-        include "com.mydomain/authelia_location.conf";
-        include "com.mydomain/authelia_location-basic.conf";
-        location / {
-                include "com.mydomain/authelia_proxy.conf";
-                include "com.mydomain/authelia_authrequest.conf";
-                root /home/web/drive;
-        }
-        location = /common {
-                 return 301 https://$host/common/;
-        }
-        location /common/ {
-                include "com.mydomain/authelia_proxy.conf";
-                include "com.mydomain/authelia_authrequest.conf";
-                client_max_body_size 512M;
-                proxy_pass http://127.0.0.1:3002;
-                proxy_set_header Connection $http_connection;
-                proxy_set_header Connection 'upgrade';
-                proxy_cache_bypass $http_upgrade;
-       }
-       location /webdav/common {
-                include "com.mydomain/authelia_proxy.conf";
-                include "com.mydomain/authelia_authrequest-basic.conf";
-                # https://mailman.nginx.org/pipermail/nginx/2007-January/000504.html - fix Destination: header
-                # https://trac.nginx.org/nginx/ticket/348 - bug, workaround with named capture
-                set $dest $http_destination;
-                if ($http_destination ~ "^https://(?<myvar>(.+))") {
-                        set $dest http://$myvar;
-                }
-                # Warning: adding / at the end of the proxy_pass will break WebDAV!
-                proxy_pass http://127.0.0.1:10001;
-                proxy_buffering off;
-                gzip off;
-                proxy_pass_request_headers on;
-               proxy_set_header Destination       $dest;
-        }
-        client_max_body_size 100M;
-}
-</file>
-This example also shows how i have integrated [[selfhost:sso|SSO Authentication]] with the filesever.
-refer to the [[selfhost:nginx|The Reverse Proxy concept]] page to activate this specific NGIX configuration. Of course you need to create the Let's Encrypt certificates and the subdomain in your DNS provider.
-==== Main Directory Page ====
-As you can spot from the above NGINX configuration, i have defined a common landing on **https://drive.mydomain.com** to provide a nice page to access the individual shares.
-For this i am using my [[services:dashboards|Simple Dashboard]] with the following **site.json**:
-<file - site.json>
-{
-    "title" : "My Drive Title",
-    "header" : {
-        "img" : "",
-        "text" : "My Drive"
-        },
-    "content" : [
-        {
-        "foldable": false,
-        "title": "",
-        "content":
-            [ {
-                "img" : "images/folder.png",
-                "text" : "Common",
-                "link" : "/common/",
-                "style" : "box-inline",
-                "new_page" : true
-            } ]
-        }
-        ],
-     "footer" : {
-        "img" : "",
-        "text" : "back home",
-        "style" : "footer-light",
-        "link" : "https://home.mydomain.com"
-    }
-}
-</file>
-===== Experimental stuff =====
-Just some additional experiments i did, for future references.
-=== Nephele-Serve ===
-Replacing WebDAV with Nephele-Serve (which will support also CardDAV/CalDAV in the future)
-https://www.npmjs.com/package/nephele-serve
-https://github.com/sciactive/nephele
-NPM needs to be enabled for the fileserver user:
-<code>
-NPM_PACKAGES="$HOME/.npm-packages"
-mkdir -p "$NPM_PACKAGES"
-echo "prefix = $NPM_PACKAGES" >> ~/.npmrc
-</code>
-And in **~/.bashrc**:
-<code>
-# NPM packages in homedir
-NPM_PACKAGES="$HOME/.npm-packages"
-# Tell our environment about user-installed node tools
-PATH="$NPM_PACKAGES/bin:$PATH"
-# Unset manpath so we can inherit from /etc/manpath via the `manpath` command
-unset MANPATH # delete if you already modified MANPATH elsewhere in your configuration
-MANPATH="$NPM_PACKAGES/share/man:$(manpath)"
-# Tell Node about these packages
-NODE_PATH="$NPM_PACKAGES/lib/node_modules:$NODE_PATH"
-</code>
-Install:
-<code bash>
-source ~/.bashrc
-npm install -g nephele-serve
-</code>
-Advantages: it's a simple server that supports pam_auth. In the future, it might **also** replace [[services:radicale|Radicale]] with a single service.
-Disadvantages: do not support base_url, so unable to host under /webdav even with reverse proxy.
-=== sFtpGO WebDAV / web browser ===
-Interesting [[https://github.com/drakkan/sftpgo|sFtpGo]] support both web-browser access and WebDAV from a single tool.
-You need to start it once then edit **sftpgo.json**:
-<code>
-"external_auth_hook": "/data/daemons/fileserver/login.sh",
-"webdavd": {
-    "bindings": [
-      {
-        "port": 10001,
-        "address": "127.0.0.1",
-        "enable_https": false,
-        "certificate_file": "",
-        "certificate_key_file": "",
-        "min_tls_version": 12,
-        "client_auth_type": 0,
-        "tls_cipher_suites": [],
-        "prefix": "/webdav/common",
-        "proxy_allowed": [],
-        "client_ip_proxy_header": "",
-        "client_ip_header_depth": 0,
-        "disable_www_auth_header": false
-      }
-    ],
-</code>
-Advnatages: easier than Apache to setup, support base_url
-Disadvantages: cannot use pam_auth and cannot disable authentication. So double-auth over reverse proxy that might be annoying.

Willy's Wiki

User Tools

Differences