Troubleshooting
A drive that does not come up almost always says why. This page collects what those answers mean, starting with the number Windows returns when it refuses a mapping.
Tip
TL;DR — A red icon with Offline or Maintenance is not an error and needs nothing from you. Anything else carries either a status word from the status vocabulary or an error number from Windows. Look the number up in the table below. If it is not there, the number itself is what a search engine and the Windows event log understand, and the diagnostics report is what makes an issue answerable.
- What is not a problem
- The mapping error codes
- Symptoms without a number
- Collecting a diagnostics report
Three states look alarming and are not:
| What you see | Why it is fine |
|---|---|
| Offline | the server did not answer. NcDavTray unmaps the drive cleanly and remounts it by itself as soon as it does |
| Maintenance | the server answered and said it is in maintenance mode. Same handling |
| Waiting for a free connection slot | both logins Windows keeps for that host are taken. The mount stays in the list and connects on its own as soon as one of them is released. See Server identities |
Watching a server come back on its own is what the application is for. The only reason to act is when a drive stays down while the server is demonstrably up.
When Windows refuses to map the drive it returns a number, and NcDavTray shows it. Only the codes with a remedy of their own get a message of their own; the rest carry the number.
| Code | Windows name | What NcDavTray says | Where to look |
|---|---|---|---|
| 1219 | ERROR_SESSION_CREDENTIAL_CONFLICT |
a dialog naming the two drives that hold the logins for that server | Server identities |
| 5 | ERROR_ACCESS_DENIED |
did not accept the credentials | the credentials |
| 1326 | ERROR_LOGON_FAILURE |
did not accept the credentials | the credentials |
| 1244 | ERROR_NOT_AUTHENTICATED |
did not accept the credentials | the credentials |
| 53 | ERROR_BAD_NETPATH |
could not be reached | the server field |
| 67 | ERROR_BAD_NET_NAME |
could not be reached | the server field |
| 31 | ERROR_GEN_FAILURE |
unexpected network error | the diagnostics report |
| 59 | ERROR_UNEXP_NET_ERR |
unexpected network error | the diagnostics report |
| 85 | ERROR_ALREADY_ASSIGNED |
nothing | nothing to do, see below |
| anything else | the number, as it came from Windows | the number is what the event log and a search engine speak |
The server was reached and turned the login down. For an account mount the usual cause is the wrong kind of password: an account mount needs an app password, not the password you log in with — see Accounts and app passwords. A mount that already has one stored shows its password field rather than the two login buttons; Clear beside the field drops the stored password and brings the browser login back. For a share link it is the share's own password, which is a different thing entirely — see Share links.
An app password that was revoked on the server gives the same answer. Nextcloud lists every one of them with its last activity under Settings → Security → Devices & sessions, so a mount that worked until recently is quickest to check there.
Windows could not resolve or reach the path at all. The server field of a mount holds a host name and nothing else — no https://, no port, no path. A full URL pasted in there produces exactly this. Only https on the standard port 443 works, so a server behind a different port cannot be mounted.
A server that is genuinely back can still look unreachable for a while, because Windows caches the fact that it did not answer. The lifetime of that cache is adjustable, see WebClient service.
The server answered, but the redirector could not make a WebDAV session out of what came back. These two are the codes without a remedy of their own, and the ones worth collecting a diagnostics report for: what Windows saw is the only thing that narrows it down.
ERROR_ALREADY_ASSIGNED means the drive letter is still taken. Right after a mapping was taken down that is not a failure but a matter of waiting: NcDavTray releases the letter, waits for Windows to let go of it, and maps again. Only if that second attempt fails does anything reach you.
| Symptom | What it is |
|---|---|
| The application icon says needs setup (service deactivated) | the WebClient service is disabled and cannot start on demand, so no WebDAV drive works. See WebClient service |
| Drive {letter} is already in use while adding a mount | another mount or another program holds that letter. Pick a different one |
| A share link is refused with a message rather than a number | the server, the link or the password. The messages are listed on Share links |
| {server} did not start a browser login | the address answered, but not with a login of its own. A mistyped host or something in front of the server that swallows the request. Enter the user name and an app password by hand instead — see Accounts and app passwords |
| The login was not confirmed in time | the browser login waits twenty minutes for you to confirm it, and that has run out. Nothing is broken; start it again |
| The app password … could not be withdrawn | removing the last mount of an account asked the server to withdraw its app password, and the server did not confirm it. The password is still valid and still listed under Settings → Security → Devices & sessions, where it can be deleted by hand |
| Large files fail, or a big folder behaves erratically in Explorer | redirector limits, not NcDavTray. See Speed and large files |
| A drive works but feels slow with big files | the difference between a drive and a synced folder. See Speed and large files |
| A mount shows No drive letter | the entry has no usable letter configured. Edit it in the Shares tab |
The repository ships a script that writes the relevant Windows state into one text file: copy NcDavTray_diag.ps1 and Run_NcDavTray_diag.cmd from tools/diagnostics/ onto the affected machine, start the .cmd and answer the prompts. It writes NcDavTray_diag_<date>_<time>.txt next to itself, and paste the full file into the issue inside a fenced code block.
Important
For a report you are going to post publicly, pick the profile Public issue / GitHub. It replaces host names, user names, labels and folder names with placeholders. The other profile keeps them, which is what you want when you are only reading it yourself.
Diagnostics report describes what each section of the file means, what is masked, and what never reaches the report at all.
