English
Troubleshooting
Find the code or the words on your screen in the section named after the place they appear; each row gives the cause and the fix. In the tables, <hub> stands for the address of the machine that runs the hub.
The panel is unreachable
| Symptom | Cause | Fix |
|---|---|---|
the browser reaches nothing at http://<hub>:8080 | the panel's unit is down, the port moved, or this network is not exposed | run systemctl status neutrino_hub_web on the hub, then read Panel ports and Exposure on Network |
http://<hub>:8080 moves to https://<hub>, which reaches nothing | HTTPS is on, so the HTTP port sends every browser to the HTTPS port, and that port is unreachable from here | read HTTPS port under Panel ports on Network, and check that the port is open on the way to the hub |
| the session ends after a panel port changed | the panel restarted on the new port, and the session cookie is named after the port, neutrino_session_<port> | sign in again at the new address |
| the session ends after Enable HTTPS or Disable HTTPS | the switch ends the session, and the port you left deletes the cookie when it sends you on | sign in again at the address the page moved to |
after Disable HTTPS, signing in over http:// comes back to the login card | the browser kept the Secure cookie from the HTTPS port and keeps it over the new one | select Clear the old session on the login card, or open the https:// address one time, which sends you back with the cookie removed |
The browser warns about the certificate
These rows apply to the HTTPS port. The hub's own certificate authority signs its certificate, and a browser trusts that authority only after it is installed there and the browser restarts.
| Symptom | Cause | Fix |
|---|---|---|
| the browser says the certificate authority is not trusted | the authority is not installed on this device | open http://<hub>:8080, select Install certificate under Settings > HTTPS, follow the steps for this device, and restart the browser |
| Enable HTTPS is greyed and the line beside it reads Install the certificate, then restart the browser. | this browser rejected the HTTPS port's certificate | install the authority for this system, restart the browser, and open the http:// address again |
| Chrome shows Not secure after the authority was imported | Chrome was told to proceed past a certificate warning before the import | restart Chrome |
| Firefox warns and Chrome on the same machine does not | Firefox keeps its own list of authorities | follow the Linux, Firefox steps under Install certificate |
| every browser warns again after a change on Settings | Regenerate certificate made a new authority | install the new authority on each device and restart the browser; http://<hub>:8080/api/hub/setting/https/authority serves it while HTTPS is on |
| the browser says the certificate is for another name | the address in the browser is missing from Certificate names, as a public address always is | open the panel at a name or address listed there, such as hub.neutrino.internal on the LAN |
Locked out
| Symptom | Cause | Fix |
|---|---|---|
| Locked after repeated failures. | too many wrong passwords | run sudo nhub unlock on the hub; it also lifts every fail2ban SSH ban |
| Wrong password. and it is forgotten | the panel password is lost | run sudo nhub reset password on the hub; every session is signed out |
| the vault passphrase is lost | the running hub still reads the vault with its own working key; a backup taken under the lost passphrase stays sealed | run sudo nhub vault rekey on the hub to set a new passphrase, then download a new backup |
vault_locked, on a client's row or a module, or Vault locked on the SSH Relay card | the box has no working copy of the vault's data key, so nothing stored in the vault opens | restore a backup under Settings with its vault passphrase; the restore writes the working copy again |
The Credentials page rejects a key
| Symptom | Cause | Fix |
|---|---|---|
key_is_public | the pasted key is a public key | paste the private key, with its BEGIN and END lines |
key_passphrase_needed | the private key is encrypted | fill Key passphrase with the key's passphrase, then save again |
Restoring a backup fails
With any of these codes, nothing is restored.
| Symptom | Cause | Fix |
|---|---|---|
backup_wrong_extension | the file is not a backup this panel writes | pick the archive that Download backup saved |
backup_corrupt | the archive is damaged | download the backup again, or pick an older one |
vault_passphrase_wrong | the passphrase does not open this backup | type the passphrase that was set when the backup was taken |
backup_mode_unavailable | the backup sets a mode this hub does not offer, such as a side gateway on the mainland edition | restore it on a hub of the full edition, on Linux |
| the restore's apply stops short | the files are restored, and one step of the apply failed | fix the cause the dialog names, then run sudo nhub apply on the hub |
A device is offline
| Symptom | Cause | Fix |
|---|---|---|
agent_offline, in the panel or on a client's row | the machine's agent is not connected to the hub | on the machine, run sudo nagent status; when it reads the service as stopped, run sudo nagent start |
| the machine is listed and its modules cannot be changed | an offline machine rejects every change, and nothing is queued | bring the machine back, then apply again |
A program is turned away by the hub
An agent shows these codes in nagent status. A client shows them as words on its hub row, listed under The state line of a hub row.
| Symptom | Cause | Fix |
|---|---|---|
protocol_too_old | the program is a 0.3 or 0.4 package, which speaks an older protocol than a 0.5.2 hub accepts | install the 0.5.2 package on that machine, which nagent --version confirms, and join with a new link. A 0.4 hub cannot be upgraded to 0.5.2: remove it and install 0.5.2 fresh, as Settings describes |
protocol_too_new | the program speaks a newer protocol than the hub | update the hub under Settings; the binding stays and the program connects again |
hub_untrusted | the hub was reset or installed again, and its certificate changed | join again with a fresh link, from Add by link on Devices for an agent and from Clients for a client |
binding_unknown | the machine's row was removed on the hub's Devices or Clients page, and the program deleted its binding | join again with a new link |
Joining a hub fails
| Symptom | Cause | Fix |
|---|---|---|
link_not_for_client or link_not_for_agent | the link is for the other kind of program: Clients makes client links, Devices agent links | make a link on the right page |
link_unreadable | the pasted link is cut off | copy the whole link again |
ticket_spent | the link is used, older than 30 minutes, or replaced by a newer link of the same kind | on a client, select Leave on the row; then join with a new link |
admission_paused | the hub paused joins after too many failed ones | leave the client running; it joins again after the seconds the code names |
unsupported_remote_install, in the Install agent dialog on Devices | the machine reports a system other than Linux, and the SSH installer is for Linux | install the agent on the machine itself and join it with a link, as Install an agent describes |
The state line of a hub row
The client window and the Android app show these words under a hub's name. A countdown runs live, and at 0 the line reads Connecting… again.
| Symptom | Cause | Fix |
|---|---|---|
| Connecting… for a long time | the client is dialling every address of the hub, and a dial waits out its connect time | wait for the round to end; the line then names a reason |
| The hub did not answer · retrying in 5 s | no address of the hub answered, and the device is not on the hub's virtual network | check that the hub runs and that the device reaches one way in, then select refresh |
| The hub is not on the virtual network · retrying in 5 s | no address answered while the device is on the hub's virtual network | check the hub's badge for that network on its Access page |
| No network | the device has no network at all | connect the device to a network; the client dials when one comes up |
| Certificate mismatch · retrying in 60 s | the certificate at the hub's address differs from the one the link pinned | after a reset or a new install of the hub, select Leave and join with a new link |
| The hub pauses new devices · retrying in 60 s | the hub paused joins after too many failed ones | leave the client running; it joins again when the countdown ends |
| The hub does not know this device | the hub's Clients page no longer lists this client | select Leave, the row's only button, then join with a new link |
| Version too old | the client and the hub speak different protocol numbers | update the older of the two, then select refresh |
| Join refused · and the reason | the hub rejected the link's ticket, as for a link used already, expired, or replaced by a newer link of the same kind | select Leave, then join with a new link from Clients |
| Replaced by another client · Reconnect | another client connected to the hub as this one | select Reconnect to take the hub back |
| Disabled by the hub | the client is switched off on Clients, and every action returns client_disabled | select Enable on its row there; the line changes by itself |
The client does not connect
| Symptom | Cause | Fix |
|---|---|---|
hub_unreachable | port 8443 on the hub is unreachable from this computer | check the network, the ways in on Access, and Exposure on the hub |
gui_webkitgtk_missing | WebKitGTK is absent on Linux | install the packages the message names, then start the client again |
gui_webview2_missing | WebView2 is absent on Windows | install the runtime the message names, then start the client again |
root_refused | the client was started with sudo | start it from your own account |
client_held | another account's client runs on this computer | quit the client in that account, or sign that account out, then start yours |
The virtual network does not connect
These show on a hub row's Virtual network line, in the desktop client and in the Android app. The client tries a failed connect only when you select Connect again.
| Symptom | Cause | Fix |
|---|---|---|
| Connecting… with This machine is registered with the console. Attach it to a network there. | in console mode, the console has not attached the device to a network yet | attach the device to the hub's network in the EasyTier console, as EasyTier describes |
overlay_join_failed, with the engine's own words after it | NetBird or EasyTier rejected the join; with NetBird, the hub often holds a one-time setup key already spent | create a reusable key in the NetBird console, paste it with Replace on the NetBird card, then select Connect again, as NetBird describes |
overlay_other_network | this device is on another NetBird network or on another hub's virtual network | on the other hub's row, select Disconnect on its virtual network line, then connect this one |
overlay_daemon_down | the NetBird or EasyTier service on this computer is stopped | reinstall the client |
bundle_missing | this install has no NetBird or EasyTier | reinstall the client |
overlay_no_address | the engine gave the device no address within 90 seconds | check the device's own network, then select Connect again |
overlay_console_invalid | EasyTier cannot use the console address the hub named | correct the console address on the hub's EasyTier card and apply |
overlay_withdrawn | the hub no longer publishes that network to this client | turn the network on again on the hub's Access page, or add the virtual network to the client's permission on Clients |
Access
| Symptom | Cause | Fix |
|---|---|---|
turning an engine on, or saving an EasyTier address, is rejected with overlay_subnet_overlap | the engine's network overlaps the other engine's or a network the hub holds an address on | change EasyTier's This box's address or the LAN's address on Network; NetBird's network is always 100.64.0.0/10 |
the Access page shows overlay_default_route_refused in red | a peer or the console gave the hub a default route through the network, and the hub deleted it | remove the exit route for the hub in the NetBird or EasyTier console; the hub keeps its own uplink as the way out |
overlay_route_overlap with The hub cannot take it away | a route on the network overlaps one of the hub's networks, and the hub could not deselect it | change or remove that route in the NetBird or EasyTier console |
direct_host_invalid | Public address under the Direct card is neither a host name nor an IP address | type a host name or an IP address, then select Apply access |
easytier_config_server_invalid | the pasted text is neither a console address nor a token | paste the whole address after --config-server, of the form tcp://host:22020/token |
easytier_invalid | Manual bootstrap peers is chosen and no bootstrap peer is listed | add at least one bootstrap peer, then apply again |
| the hub does not respond on a virtual network | that network is not in Exposure | tick it under Exposure on Network |
| NetBird's badge reads not joined | the hub has no NetBird identity, or its last login expired | join again with a new setup key |
| NetBird's badge reads management unreachable | the hub cannot reach NetBird's management server | on the hub, check the uplink, that DNS resolves the management server's name, and the management URL |
| NetBird's badge reads not running | NetBird is stopped | turn the card on and select Apply access |
| NetBird's Peers is empty | no other device has joined the network | on a client, select Connect on the hub row's virtual network line |
| a client reaches the hub over NetBird and the hub's LAN devices stay unreachable | the console has no access policy for the LAN route | add the network and its policy in the NetBird console, as LAN devices over NetBird describes |
| EasyTier reads No machine has joined yet. | the other machine's network name or secret differs | copy the command again with Copy with the secret |
| a device joins over EasyTier and the hub's LAN devices stay unreachable | the LAN subnet is not routed on the network | in manual mode, add the subnet under Exported networks and apply; in console mode, add it as a subnet proxy in the console, as LAN devices over EasyTier describes |
| the SSH Relay card's Status reads Not configured | Server, Account, or the key or login is empty or gone | fill in the settings and select Apply SSH Relay |
| Authentication failed | the server rejected the key or the password | check the key's line in the account's authorized_keys, or the password and PasswordAuthentication |
| Forward refused | the server refused the listener, or another program holds the port | check that permitlisten names the Public port, and free the port on the server |
| Public port closed | the forward runs, and the public address returns no answer or another certificate | check GatewayPorts clientspecified, the provider's firewall and the server's firewall |
| Host key changed | the server presents a key other than the one recorded | if you replaced or reinstalled the server, select Forget host key |
| Server unreachable | SSH could not reach the server, or the connection dropped | check Server and SSH port, and that the server is up |
Network settings are refused
| Symptom | Cause | Fix |
|---|---|---|
network_invalid with the field role: the panel reads that the port cannot answer with the role Disabled | a port ticked under Exposure has the role Disabled | give the port the role WAN or LAN, or untick it under Exposure, then apply again |
Country code reads Two letters, such as DE. 5 GHz needs one., or network_invalid with the field country_code | an access point on 5 GHz has no valid two-letter country code | type the code of the country the hub is in under Country code, such as DE, then apply again |
port_out_of_range, on the Network, Access, Proxy or Settings page | a port field holds a number outside 1 to 65535 | type a port from 1 to 65535, then apply again |
port_already_in_use | another program on the hub's computer already listens on that port | pick another port, or stop the program that holds it |
resolver_required, on the Proxy page | the Remote resolvers (proxied names) list is empty | add at least one resolver, then select Apply route |
Errors on the Modules page
A tab of the Modules page shows these codes under its state word or on a row, and a refused apply names one.
Every module
| Symptom | Cause | Fix |
|---|---|---|
module_fetch_failed, with the download's own words after it | the hub could not download the module's install file | check that the hub reaches the internet, then select Install again |
a module is greyed with This machine's system cannot run it, or no_platform_build | the module has no build for this machine's system | pick another machine; Supported platforms lists each module's systems |
gitea_git_missing | a Mac or a Windows machine has no usable git for Gitea | install git, then select Install again |
Instances of VS Code, code-server, CloudCLI and Codeg
These codes also come from the Terminal tab and from the accounts under Global configuration.
| Symptom | Cause | Fix |
|---|---|---|
account_unknown | the machine has no account by that name | type the name of an account that exists on the machine, or create it there |
account_invalid | the name cannot be an account name on that machine | type a valid account name |
account_duplicate | two instances name the same account | remove one of them |
port_duplicate | two instances use the same port | give each instance its own Port |
port_invalid | the port is outside 1024 to 65535 | type a port in that range |
credential_missing | a Windows instance has no login picked | pick the account's login under Windows login for the account |
credential_invalid | Windows no longer accepts the instance's login, as after a password change | add a login with the new password on Credentials, pick it, and apply again |
token_missing | the hub sent no connection token for the instance | apply the module again |
secret_missing | the hub sent no token secret for a code-server or Codeg instance | apply the module again |
CloudCLI
| Symptom | Cause | Fix |
|---|---|---|
cloudcli_node_download_failed | Node.js for CloudCLI could not be fetched or unpacked | check that the machine reaches the internet, then select Apply CloudCLI |
cloudcli_npm_install_failed | npm failed to install CloudCLI for the account; the row shows npm's last lines | fix what npm's lines name, then select Apply CloudCLI |
cloudcli_native_module_failed | better-sqlite3, node-pty or bcrypt could not fetch its prebuilt binary | check the machine's internet access, then select Apply CloudCLI |
cloudcli_install_out_of_memory | the machine ran out of memory during the npm install | free memory on the machine, then select Apply CloudCLI |
cloudcli_port_taken | another program on the machine holds the instance's port | change the instance's Port, then apply again |
cloudcli_register_failed | the account's CloudCLI already has an administrator with another password | wait: the agent starts that CloudCLI afresh with a new sign-in file, then signs in again every 30 seconds until it succeeds |
Codeg
| Symptom | Cause | Fix |
|---|---|---|
codeg_download_failed | the server bundle or the Node.js archive did not reach the machine from the hub, or did not unpack | check the hub's own download first, then select Install or Apply Codeg again |
codeg_port_taken | another program on the machine holds the instance's port | change the instance's Port, then select Apply Codeg |
VS Code
| Symptom | Cause | Fix |
|---|---|---|
terms_not_accepted | nobody accepted Microsoft's terms for VS Code on this machine | select Open and accept the terms on the VS Code tab |
code-server
| Symptom | Cause | Fix |
|---|---|---|
code_server_download_failed | the release could not be fetched or unpacked | check that the machine reaches the internet, then apply again |
code_server_port_taken | another program on the machine holds the instance's port | change the instance's Port and select Apply code-server again |
Remote desktop
The tab, the device's drawer on Devices, and a client's Connect show these codes. Apply remote desktop tries a failed step again.
| Symptom | Cause | Fix |
|---|---|---|
rdp_takeover_failed | a step of starting the agent's RustDesk failed; the code names the step | fix what the step names, then select Apply remote desktop |
rdp_restore_failed | a step of putting back the machine's own RustDesk failed | fix what the step names, then select Apply remote desktop |
rdp_nobody_seated | nobody is signed in at the machine's screen | sign in at that machine's own screen |
rdp_screen_not_allowed | a Wayland session has not allowed screen sharing | allow screen sharing one time at that machine's own screen |
rdp_permissions_needed | the Mac has not given RustDesk Screen Recording and Accessibility | grant both to RustDesk in that Mac's System Settings > Privacy & Security |
Terminal
| Symptom | Cause | Fix |
|---|---|---|
path_invalid | the shell program is not a full path | type the program's full path, or pick it with Browse… |
shell_program_unusable | the program is missing on the machine, or it cannot run; for nclient terminal exec, the command's program | pick a program that exists and runs, or clear Shell program; for exec, check the command's name or path |
AI tools
| Symptom | Cause | Fix |
|---|---|---|
switch_failed | cc-switch could not point the account's tools at the gateway | read the words on the account's row, then turn the setting off and on |
cc_switch_download_failed | the machine could not get cc-switch from the hub | check that the hub reaches the internet, then turn the setting off and on |
| cc-switch did not put the tools back to their own settings: and cc-switch's words | the switch back failed, and the control stays on | fix what cc-switch printed, then select the control again |
A client page cannot reach its service
The hub or the agent of the providing machine returns these codes, and the client shows them on the entry's row.
| Symptom | Cause | Fix |
|---|---|---|
connect_failed with refused | nothing listens on the service's port on that machine | start the service, or the module, on that machine |
connect_failed with timeout | the service did not respond within 10 seconds | check that the machine and the service are up |
connect_failed with unreachable | the hub has no route to a service declared by hand on Services | check the declared address, and that the hub's LAN reaches it |
port_not_published | the machine stopped publishing the port: the container, the instance or the share stopped | start it again on the Modules page |
connect_limit | the client has 256 connections open through the hub | close forwards or programs the client does not need |
permission_denied | the client's permission on Clients leaves out that kind of entry, that machine, terminals, remote commands or the panel | widen the client's permission on Clients |
service_unknown | the hub no longer publishes the entry | select refresh; the entry leaves the list |
port_taken | another program on this computer listens on the entry's Fixed local port | select Configure on the row and pick another port |
web_token_missing | the hub sent no token for a VS Code, code-server or CloudCLI entry | select Open again |
A share does not mount on a computer
| Symptom | Cause | Fix |
|---|---|---|
mountpoint_invalid | the path is outside your home | give a folder under your home, such as ~/nas/media |
mountpoint_not_drive_letter | Windows mounts a share at a drive letter | pick an unused letter, such as N: |
mountpoint_not_empty | the folder holds files | pick an empty folder |
mountpoint_in_use | another entry, from any hub, mounts at that path | pick another folder or drive |
mount_not_authorized | the polkit prompt or the macOS login dialog was dismissed. On Linux, polkit rejects the mount without a prompt when the client was started outside a desktop session, such as from an SSH login | select Mount again and confirm the prompt. Outside a desktop session, quit the client and start it from the desktop, or with systemd-run --user nclient gui |
mount_tooling_missing | mount.cifs or the client's mount helper is absent | install the client package again, which depends on cifs-utils |
pkexec_missing | the computer has no pkexec, which a mount on Linux goes through | install the pkexec package, then select Mount again |
mount_timed_out | on macOS, the system did not finish the mount within ten minutes | select Mount again |
credentials_missing | the saved credentials file is gone | select Configure and type the password again |
share_login_rejected | the share rejected the username or the password | select Configure and enter both again |
share_access_denied | the share accepts the login and denies that account access | pick the user under Who may use it for that share on the Modules page |
share_not_found | the host has no share by that name | check the share on the Modules page or on Services |
share_session_conflict | Windows holds a connection to that server under another login | disconnect that connection in Windows, then select Mount again |
files_adapter_unavailable on Windows | the NeutrinoClientFiles service is missing, stopped or failing, so the files adapter could not start | read the words after the code, run sc query NeutrinoClientFiles, and install the client again if the service is missing |
files_adapter_in_use on Windows | another account's client holds the files adapter | quit the client in that account, or sign that account out; the mount runs again after that |
| the row is greyed, Not reachable now | the serving machine is off or not connected to the hub | bring the machine back |
A computer shows share_unreachable for the same causes as a phone, listed in the next section.
A share is unreachable on a phone
The Files app on the phone shows the Android app's shares, and reports a failure as the code's sentence.
| Symptom | Cause | Fix |
|---|---|---|
share_unreachable, with the hub's row not Connected | the device cannot reach the hub's port 8443 | bring the hub's row to Connected through any way in on Access: the LAN, NetBird, EasyTier or the SSH Relay |
share_unreachable while the hub's row reads Connected | the machine that serves the share is off or not connected to the hub | bring the machine back |
| Files reads Give this share's password on the Files screen first. | the share has no login on this phone yet | open the share on the app's Files screen, enter the username and password, and select Connect |
A terminal closes
| Symptom | Cause | Fix |
|---|---|---|
session_not_owned | the session belongs to another viewer and is not shared, or persist or share was asked by a client that did not open it | ask its owner to switch on Shared, or open a terminal of your own |
session_unknown | the session ended, or the agent restarted or updated, which ends every session on that machine, or the id given to attach, persist, share or stop matches no session | open a new terminal on the machine |
shell_unknown | the hub no longer holds the shell the client resized | close the terminal in the client and open it again |
unknown_terminal | the hub offers no terminal on that machine, or nclient terminal named an unknown machine | pick a machine the client's Terminals page lists |
A remote desktop does not open
A code about the sharing machine's screen, such as rdp_nobody_seated, is under Remote desktop.
| Symptom | Cause | Fix |
|---|---|---|
| the page reads No shared remote desktops yet | no machine is sharing, or the sharing machine is offline | turn on Share this machine's desktop on that machine's tab |
rdp_viewer_open | a viewer is already open on that desktop from this client | close the open viewer, then select Connect again |
rdp_not_shared | the machine stopped sharing | turn the share on again on that machine's Remote desktop tab |
rdp_no_desktop | this session has no screen to open a viewer on | connect from a desktop session |
rdp_launch_failed | the viewer did not start, with the reason after the code | fix the reason, then select Connect again |
An AI tool ignores the gateway
| Symptom | Cause | Fix |
|---|---|---|
no_endpoint | the hub has issued this client no key | on the hub's AI page, revoke the client's key so a new one is issued, or join the client again |
no_exit_hub | the switch was turned on for a hub that is not connected | wait until the hub's row reads Connected, then turn the switch on |
| the tools keep their own configuration after Save in AI tool configuration | Save keeps the model choices and points nothing | switch on The AI tools use this gateway on the entry's row; nclient service ai show prints where the tools point |
the panel reads gateway_unreachable | the gateway process is down | run systemctl status neutrino_hub_cliproxyapi on the hub |
| a tool reaches nothing | the tools point at the client's forward on 127.0.0.1, which listens only while the client runs | start the client, or keep it running in the tray |
login_expired | a subscription sign-in on the AI page took too long | select Try again and finish the sign-in in the browser |
the client's AI page is greyed with ai_tools_managed's sentence | the Neutrino agent is installed on this computer, and it sets the AI tools | set the AI tools on the hub's Modules page, under Global configuration for this machine |
Where the logs are
| What | Where |
|---|---|
| the setup run | setup.log in the hub's log directory, /var/log/neutrino/hub/setup.log on Linux |
| the hub on Linux | each unit's journal: journalctl -u neutrino_hub_web -n 200, and likewise neutrino_hub_xray, neutrino_hub_dnsmasq, neutrino_hub_cliproxyapi, neutrino_hub_netbird, neutrino_hub_easytier, neutrino_hub_relay |
| the hub on macOS and Windows | one <name>.log per process in /Library/Logs/Neutrino/hub/ on macOS and C:\ProgramData\Neutrino\hub\log\ on Windows |
| a hub update | journalctl -u neutrino_hub_update on Linux; hub_update.log in the hub's log directory on macOS and Windows |
| the agent on Linux | journalctl -u neutrino_agent -n 200 on that machine |
| the agent on Windows | C:\ProgramData\Neutrino\agent\log\agent.log |
| the agent on macOS | /Library/Logs/Neutrino/agent/agent.log |
| the desktop client | client.log in ~/.config/neutrino/client/ on Linux, ~/Library/Logs/Neutrino/client/ on macOS, and %LOCALAPPDATA%\Neutrino\client\ on Windows |
| the client's EasyTier daemon | /var/log/neutrino/client/ on Linux, /Library/Logs/Neutrino/client/ on macOS, and C:\ProgramData\Neutrino\client\log\ on Windows |
| the AI gateway, from the panel | Journal on the AI page |
| what a render produces | sudo nhub apply --dry-run on the hub |