When something is not working: where to start

13 min · 12 sections

This guide is ordered by what you see, not by what is broken. Find the symptom and it tells you what to check, and in what order.

One rule first: the panel says what it knows and does not guess. If a screen gives you a reason, that reason has been checked; if it says "could not look", it really could not, and that is where this guide comes in.

1. I cannot open the panel

First separate two very different things: the NAS is not answering or your browser will not go in. Ten seconds tells them apart.

1.1 Is it still at that address?

The number one cause of "it does not open" is not a fault: the IP changed. It happens when the router restarts and hands out addresses again.

With a monitor and keyboard on the NAS, or over SSH:

lgm

It shows the version, the machine name and the panel's address. If it is different from the one you had saved, that was the problem, and the real fix is to give it a fixed address: Control Panel → Network, or lgm set-ip from the console.

1.2 Is the service running?

systemctl status nas-backend nas-helper

Those are the two that matter: nas-backend is the panel and nas-helper is the privileged process that does the system work. If either is stopped:

sudo systemctl restart nas-backend
lgm log

lgm log shows the panel's log, which is where the reason will be.

⚠️ If you have just updated and the panel does not come back, see section 6.

1.3 "The NAS is not responding right now (restarting?)"

This message appears inside the panel: the page loaded, so the browser reached the NAS at some point. Two cases:

• You just updated or restarted. Normal for a few seconds; the panel retries on its own.
• It shows immediately and never goes away. Then the page you are looking at came from the browser's saved copy, not from the NAS. It is usually the certificate: the browser refuses the connection and never gets to show you its own warning, which is the only screen where you can accept it. The message itself has an "Enter without the saved copy" button that clears that copy and reloads; then the usual certificate warning appears.

2. I get in, but an app says I have no access

That is not a fault: it is how apps are shared out. Every account has its own apps and at which level — look, or look and touch — and the NAS decides, not the icon.

Whoever administers changes it in Control Panel → Users and groups → the account → Apps. It is covered in full in Users, permissions and who sees what.

Two things that confuse people:

• The icon is gone. Removing an app also removes it from that account's desktop and search. It has not been uninstalled: it is not theirs.
• I can see it but not change it. That is the "view" level. The buttons that write are not shown. In Cameras, Backups and Storage it is the most an account that does not administer can be given.

3. I cannot see the NAS from Windows, Mac or my phone

In order, from most common to least:

1. Does that account have access to the folder? Sharing a folder and giving someone access are two different things: Control Panel → Household folders → the folder → Permissions.
2. Is the protocol on? SMB, NFS and WebDAV are switched on separately in `Control Panel → File services`, and each must also be ticked on each folder.
3. The SMB password is not the panel one if you never set it. They are set together when the account is created, but changing one does not change the other.
4. Type the address by hand instead of waiting for it to appear: on Windows \\192.168.1.50, on Mac smb://192.168.1.50. Automatic network discovery is the first thing that breaks and it means nothing.
5. Is only My Drive missing? It has its own row in Control Panel → Household folders: if it is switched off, or does not have the protocol you use, it does not show up. Over NFS, on top of that, only the machines on its list can mount it.

All the detail, device by device, in Connecting your devices.

4. It is slow

Before touching anything, look at who is working: Monitor shows CPU, memory, network and disks live. With that, three cases:

• Disks at 100 % and everything else quiet. Almost always a scrub, a backup or the photo indexer. It passes; Storage and Backups show what is running.
• Network maxed out. Someone is copying something, or a torrent is seeding. The download speed limits are in the Downloads app.
• Everything quiet and still slow. If it is a torrent, check the port warning in Downloads: with the port closed you only download from peers you reach out to, which is a fraction of the speed. If it is SMB, try copying one big file instead of many small ones: a thousand 10 KB files are a thousand conversations, and that is slow on any NAS.

5. A disk is reporting errors

Do not unplug it yet. With RAID, pulling the wrong disk turns a problem into a loss.

1. Storage → Protection → Health and encryption says which disk it is and what is wrong (SMART). Write down its **serial number**, which is the only thing that does not move; the letter (sda, sdb) can change on reboot.
2. If the volume still works, make a backup now before touching anything. A disk that starts failing usually holds on for hours.
3. Replacement is done from the panel and is covered step by step in [Disaster recovery](disaster-recovery.md).

6. I updated and something is wrong

lgm log and lgm progress tell you what happened in the last update.

• The panel does not come back. The updater keeps a copy of the units before touching them precisely for this. systemctl status nas-backend will say whether it fails to start, and lgm log why.
• A critical task half done. If a restore or a backup is running, the update refuses to start. That is on purpose.

Going back to the previous version

If a new version breaks something and you cannot wait for the fix, Control Panel → Update LGM-OS has a «Go back to X» button with the version that was there before the last update. It returns the panel to that version in about a minute; your data and settings stay as they are and apps keep running. The retired version does not reinstall itself: whenever you want it, «Install anyway». The button is not shown when the update changed the shape of the settings (the previous version would not understand them); the panel says so.

7. The disk filled up

First work out which disk: the system one and the data one are different and are fixed differently.

• The data volume. Control Panel → Space report shows where the disk is going and the duplicates. Watch out for snapshots: what you delete keeps taking up room while a snapshot still remembers it.
• The system disk. It fills up with logs, Docker images and downloads pointed at the wrong place. The Clean button in the Package Centre — which removes images and volumes nobody uses any more — and the recycle bin are the first two places to look.

8. How to ask for help with data

In Control Panel → System check there is a button, Download support report, that downloads a file with everything needed to understand what is wrong with the machine: the version, how the last update went, the service check, the disks, the network, the Docker apps, the alerts and the latest panel log lines (at most 5,000, from the last day). Preparing it can take a minute. Send it as it is to whoever is helping you; it is what saves the ten back-and-forth questions. It carries no passwords: not the NAS accounts, not the cameras, not the clouds. The fields that hold them come out as [oculto].

If you would rather look at it yourself from the console:

lgm info      # version, machine and address
lgm log       # the panel's log
systemctl status nas-backend nas-helper

And the Network tests section of the System check has the network checks — resolve a name, reach the internet, see whether a port answers — without leaving the browser.

⚠️ Before pasting a log anywhere, read it. The panel strips passwords from the messages it shows, but a system log is the system's and may carry internal addresses or your folder names.

9. "The «…» setting was damaged"

The panel's settings live in files under /var/nas/state, one per topic: users.json the accounts, shared_folders.json the household folders, and so on. If one cannot be read, the panel does not fall over: it sets it aside as <name>.json.corrupt, carries on with the factory values for that part and warns you with "The «…» setting was damaged". What you notice depends on which one:

⚠️ Until it is recovered, do not create or edit anything in that part of the panel: you would save the factory values over it. With shared_folders set aside, creating a folder would publish only that one.

The file is almost always fine. Up to 2.0.4, a read that went wrong set aside files that were perfect on disk. Since 2.0.4 it no longer happens, and on start the panel puts back by itself the ones that read fine: you will see "The «…» setting recovered by itself".

9.1 Recovering it

Over SSH, or with a monitor and keyboard plugged into the NAS:

1. On a version older than 2.0.4, update. When the new version starts it puts the file back by itself, and nothing else is needed:

sudo lgm update

1. If you already have 2.0.4 or later, restart the panel; on start it checks the files set aside:

sudo systemctl restart nas-backend

1. By hand, without updating or restarting. Paste all of this. It puts back every file set aside that is a good setting, does not overwrite one that has been saved again and leaves alone the ones that are really broken:

sudo python3 - <<'EOF'
import glob, json, os
d = "/var/nas/state"
for apartado in sorted(glob.glob(d + "/*.json.corrupt")):
    nombre = os.path.basename(apartado)[:-len(".json.corrupt")]
    destino = f"{d}/{nombre}.json"
    if os.path.exists(destino):
        print(f"- {nombre}: a new one already exists, left alone"); continue
    try:
        doc = json.loads(open(apartado, "rb").read().decode("utf-8"))
    except Exception as e:
        print(f"- {nombre}: really broken, left alone ({e})"); continue
    if not isinstance(doc, dict):
        print(f"- {nombre}: not shaped like a setting, left alone"); continue
    st = os.stat(apartado)
    tmp = destino + ".reponiendo"
    with open(tmp, "wb") as f:
        f.write(json.dumps(doc, indent=2, ensure_ascii=False).encode("utf-8")); f.flush(); os.fsync(f.fileno())
    os.chown(tmp, st.st_uid, st.st_gid); os.chmod(tmp, 0o600); os.replace(tmp, destino)
    print(f"+ {nombre}: put back")
EOF

Every line with + is a setting put back, and it shows straight away: with users, you can sign in again.

9.2 If it is really broken

If step 3 says "really broken", what is in the file cannot be used and it has to come from the configuration backup.

• You can sign in to the panel (the broken one is not users): Control Panel → Configuration backup → "Restore from file…".
• You cannot sign in (the broken one is users): take it out of the backup over SSH. First, the most recent one:

sudo find /mnt /srv -name 'nas-config-*.tar.gz' -printf '%T+ %p\n' 2>/dev/null | sort | tail -3

And from it, only that setting (change the path to the one shown above):

COPIA=/mnt/tank/backups/nas-config-XXXX.tar.gz
sudo tar -xzf "$COPIA" -C /tmp state/users.json
sudo install -o "$(stat -c %U /var/nas/state)" -g "$(stat -c %G /var/nas/state)" -m 600 \
  /tmp/state/users.json /var/nas/state/users.json
sudo rm -rf /tmp/state

• No backup of the accounts: restart the panel (sudo systemctl restart nas-backend). With no accounts, the first-start wizard comes back; it asks for the setup code, which only shows on the NAS's own screen or with sudo lgm alta. Create the administrator and add the other accounts again. Your files and folders are not touched.

10. An alert about errors on the PCIe bus

The NAS watches the PCIe bus, through which NVMe disks, disk controllers, network cards and the graphics card talk to the motherboard, and alerts you as soon as one starts reporting errors, saying which one.

• "Corrected PCIe bus errors": the system has corrected them and nothing is being lost for now. If the alert repeats and the count keeps growing, it is usually a loose contact or a device starting to fail: shut down the NAS and reseat it in its slot.
• "PCIe bus errors": errors the system could not correct. It is usually a device not seated properly, running hot or starting to fail. If it is a disk, back up what matters now; then shut down the NAS, reseat the device and check it gets airflow.

11. The panel's keys and the TPM chip

The keys the panel keeps in order to work — the VPN's, the dynamic DNS's, the UPS's and others — are stored encrypted with the motherboard's TPM 2.0 chip if the device has one. There is nothing to switch on, and on a device without a chip they are stored as always.

If one day the chip stops opening them, the NAS still starts and alerts you with "The security chip does not open the panel's secrets". It usually happens after changing the motherboard or clearing the TPM in the BIOS. Meanwhile the VPN, dynamic DNS and anything else that uses those keys will not work.

• Nothing has been lost. Sign in to the panel with an administrator account, typing its password, and the keys are recovered automatically. The "Panel secrets recovered" alert confirms it.
• If the chip works, they are stored with it again; if the device no longer has a working chip, they are stored as before, without it.

See also

• Disaster recovery — when something serious has already happened.
• Users, permissions and who sees what — most of "I cannot see this".
• Glossary — what each odd word here means.