{"id":16079,"date":"2026-10-09T16:12:22","date_gmt":"2026-10-09T14:12:22","guid":{"rendered":"https:\/\/monodes.com\/predaelli\/?p=16079"},"modified":"2026-10-09T16:12:24","modified_gmt":"2026-10-09T14:12:24","slug":"nas-with-autosuspend-wakeonlan-and-sshfs-automount","status":"publish","type":"post","link":"https:\/\/monodes.com\/predaelli\/2026\/10\/09\/nas-with-autosuspend-wakeonlan-and-sshfs-automount\/","title":{"rendered":"NAS with Autosuspend, WakeOnLAN and SSHFS Automount"},"content":{"rendered":"\n

NAS with Autosuspend, WakeOnLAN and SSHFS Automount<\/a><\/p>\n\n\n\n\n\n\n\n\n\n\n\n

\n

NAS with Autosuspend, WakeOnLAN and SSHFS Automount<\/h1>\n\n\n\n

May 30, 2020<\/p>\n\n\n\n

This post describes the setup of some convenient features that I deployed for the NAS, to automatically save power, make it wake up and mount it on the client only when needed.<\/p>\n\n\n\n

Autosuspend<\/h3>\n\n\n\n

Autosuspend is a nice tool to let a server suspend if it is idle (with configurable conditions). For Fedora there is no official package available, so I installed it from source:<\/p>\n\n\n\n

git clone https:\/\/github.com\/languitar\/autosuspend.git; cd autosuspend\npython3 setup.py install --prefix=\/usr\nmv \/usr\/local\/lib\/python3.7\/site-packages\/autosuspend-3.0.0.dev0-py3.7.egg\/etc\/* \/etc\nmv \/usr\/local\/lib\/python3.7\/site-packages\/autosuspend-3.0.0.dev0-py3.7.egg\/lib\/systemd\/system\/* \/usr\/local\/lib\/systemd\/system\/\n<\/code><\/pre>\n\n\n\n
Starting with usual git clone and python module install, unfortunately the setup does not install the config and service files, so I had to move them manually afterwards.<\/p>\n\n\n\n
Based on the default config this is resulting the \/etc\/autosuspend.conf<\/code>:<\/p>\n\n\n\n
[general]\ninterval = 30\nidle_time = 900\nsuspend_cmd = \/usr\/bin\/systemctl suspend\nwakeup_cmd = sh -c 'echo 0 > \/sys\/class\/rtc\/rtc0\/wakealarm && echo {timestamp:.0f} > \/sys\/class\/rtc\/rtc0\/wakealarm'\nwoke_up_file = \/var\/run\/autosuspend-just-woke-up\nlock_file = \/var\/lock\/autosuspend.lock\nlock_timeout = 30\n\n# This will use the Users check with the custom name RemoteUsers.\n# Custom names are necessary in case a check class is used multiple times.\n# Custom names can also be used for clarification.\n[check.RemoteUsers]\nclass = Users\nenabled = true\nname = .*\nterminal = .*\nhost = [0-9].*\n\n# Custom\n[check.SSH]\nclass = ActiveConnection\nports = 22\nenabled = true\n\n[check.NetworkBandwidth]\nenabled = true\ninterfaces = eno1\nthreshold_send = 200\nthreshold_receive = 400\n<\/code><\/pre>\n\n\n\n
Three checks are enabled here, RemoteUser checks is someone is currently logged in over ssh, the ActiveConnection check on port 22 will reliably detect SSHFS connections. Also if there are other services causing some bandwidth usage it will detect it and not suspend with the last check.<\/p>\n\n\n\n
After starting\/enabling with systemctl enable --now autosuspend.service<\/code> it should look like this in systemctl status autosuspend.service<\/code>:<\/p>\n\n\n\n
\u25cf autosuspend.service - A daemon to suspend your server in case of inactivity\n   Loaded: loaded (\/etc\/systemd\/system\/autosuspend.service; disabled; vendor preset: disabled)\n   Active: active (running) since Fri 2020-05-15 12:23:19 CEST; 8h ago\n     Docs: https:\/\/autosuspend.readthedocs.io\/en\/latest\/systemd_integration.html\n Main PID: 1348 (autosuspend)\n    Tasks: 1 (limit: 4536)\n   Memory: 15.4M\n   CGroup: \/system.slice\/autosuspend.service\n           \u2514\u25001348 \/usr\/bin\/python3 \/usr\/bin\/autosuspend -l \/etc\/autosuspend-logging.conf daemon\n\nMay 15 20:50:50 nas-server autosuspend[1348]: 2020-05-15 20:50:50,740 - autosuspend.Processor - INFO - Starting new check iteration\nMay 15 20:50:50 nas-server autosuspend[1348]: 2020-05-15 20:50:50,752 - autosuspend.Processor - INFO - Check SSH matched. Reason: Ports [22] are connected\nMay 15 20:50:50 nas-server autosuspend[1348]: 2020-05-15 20:50:50,753 - autosuspend.Processor - INFO - System is active. Resetting state\n<\/code><\/pre>\n\n\n\n
Here we see that it currently detects the SSH connection and does not suspend when I am logged in.<\/p>\n\n\n\n
Wake On LAN<\/h3>\n\n\n\n
First make sure WOL is set up on the server and works from the client with wol xx:xx:xx:xx:xx:xx<\/code> using the MAC address of the server. This usually has to be enabled in the BIOS and might also need some software, read more here<\/a>.<\/p>\n\n\n\n
Now we want the client to wake up the server automatically. I searched for some guidance and found this blog post<\/a>, he used a custom socket that triggers WOL on connection. It works, but I observed reduced performance with this approach: instead of >900 Mbit\/s it delivered only about 300 Mbit\/s, because all data must go through netcat and the local socket.<\/p>\n\n\n\n
So I came up with a solution where the data does not get intercepted and therefore does not impact the performance. For that an extra systemd service is running to check if the server is already up and if not sends the WOL packet. For example on ArchLinux custom services should be placed into \/etc\/systemd\/system<\/code>, I created there a nas-online.service<\/code>:<\/p>\n\n\n\n
[Unit]\nDescription=Continuously check if the host is up, otherwise try WOL on start\nAfter=network.target network-online.target\nPartOf=mnt.mount\n\n[Service]\nEnvironment=HOST=nas-server\nEnvironment=PORT=22\nEnvironment=MAC=xx:xx:xx:xx:xx:xx\n\nExecStartPre=\/bin\/sh -c 'for n in `seq 1 6`; do nc -z ${HOST} ${PORT} && break || wol $MAC && sleep 10; done'\nExecStart=\/bin\/sh -c 'while :; do (nc -z -w 1s ${HOST} ${PORT} && echo \"Host is up!\" && sleep 30) || break; done'\n<\/code><\/pre>\n\n\n\n
In the [Unit] section it is defined as PartOf mnt.mount<\/code> (which we will create in the next section), this makes systemd propagate start\/stop commands from the mount to this unit. The [Service] contains 2 small shell scripts. Pre start runs a maximum of 6 times a loop where it checks with nc -z<\/code> (netcat zero IO) if the target is up, otherwise it sends the WOL packet and waits 10 secs. Then the main script just check every 30 secs if the target is still online and logs it. Per default this runs indefinitely, but as described above it does get stopped if mnt.mount<\/code> gets inactive.<\/p>\n\n\n\n
Test it with systemctl start nas-online.service<\/code>, systemctl stop nas-online.service<\/code> and systemctl status nas-online.service<\/code>:<\/p>\n\n\n\n
* nas-online.service - Continuously check if host is up, otherwise try WOL on start\n     Loaded: loaded (\/etc\/systemd\/system\/nas-online.service; static; vendor preset: disabled)\n     Active: inactive (dead) since Sat 2020-05-30 16:06:16 CEST; 4min 3s ago\n    Process: 3589 ExecStartPre=\/bin\/sh -c for n in `seq 1 6`; do nc -z ${HOST} ${PORT} && break || wol $MAC && sleep 10; done (code=exited, status=0\/SUCCESS)\n    Process: 3768 ExecStart=\/bin\/sh -c while :; do (nc -z -w 1s ${HOST} ${PORT} && echo \"Host is up!\" && sleep 30) || break; done (code=killed, signal=TERM)\n   Main PID: 3768 (code=killed, signal=TERM)\n\nMay 30 16:05:01 pc-arch systemd[1]: Starting Continuously check if host is up, otherwise try WOL on start...\nMay 30 16:05:04 pc-arch sh[3617]: Waking up xx:xx:xx:xx:xx:xx...\nMay 30 16:05:14 pc-arch systemd[1]: Started Continuously check if host is up, otherwise try WOL on start.\nMay 30 16:05:14 pc-arch sh[3774]: Host is up!\nMay 30 16:05:44 pc-arch sh[4140]: Host is up!\nMay 30 16:06:15 pc-arch sh[4198]: Host is up!\nMay 30 16:06:16 pc-arch systemd[1]: Stopping Continuously check if host is up, otherwise try WOL on start...\nMay 30 16:06:16 pc-arch systemd[1]: nas-online.service: Succeeded.\nMay 30 16:06:16 pc-arch systemd[1]: Stopped Continuously check if host is up, otherwise try WOL on start.\n<\/code><\/pre>\n\n\n\n
Automount<\/h3>\n\n\n\n
The Systemd automount feature can automatically mount it when an access to the mountpoint is detected and unmount if it was idle for a specified time.<\/p>\n\n\n\n
First it is important to login as root with SSH and accept the fingerprint for the server, otherwise it will not work because it is not in the known hosts list (like describe here<\/a>):<\/p>\n\n\n\n
sudo -i\nssh -p 2222 jk@nas-server\n<\/code><\/pre>\n\n\n\n
Also it is of course not possible to enter a password, so make sure pubkey authentication is properly configured on the server.<\/p>\n\n\n\n
Then an mount entry in \/etc\/fstab<\/code> can be added:<\/p>\n\n\n\n
jk@nas-server:\/mnt   \/mnt    fuse.sshfs  noauto,x-systemd.requires=nas-online.service,identityfile=\/home\/jk\/.ssh\/nas_ed25519,user,allow_other,default_permissions  0 0\n<\/code><\/pre>\n\n\n\n
Here we need a few options:<\/p>\n\n\n\n