Skip to content

Est. 2011ยทMicrosoft Partner ID 7033487ยทDelivery under 3 minยทSupport 7 days a week

Your vault is empty.

Free Fix 17182

Error 17182: TDSSNIClient Initialisation Failed and the Port Never Opens

11 min read Updated October 5, 2026 SQL Server

Fix it now

SQL Server could not initialise its network endpoints, so it stopped rather than run unreachable. The 17182 line prints an error value, a status code and a Reason, and the Reason is the diagnosis. Errors 17826 and 17120 that follow are consequences, not separate problems.

Run in an elevated Command Prompt on the SQL Server itself

netstat -ano | findstr :1433
certutil -store my > %TEMP%\cert.txt
  1. Open the current error log. Its folder is the -e startup parameter on the instance’s Startup Parameters tab in SQL Server Configuration Manager; the file is ERRORLOG with no extension.
  2. Find the 17182 line and read all of it: an error value, a status code and a Reason in words. “All protocols are disabled” appears as error 0xd with status code 0x4, and Windows reports it as service-specific error 13.
  3. For that case, open SQL Server Configuration Manager, SQL Server Network Configuration, Protocols for your instance, and enable at least Shared Memory. Restart the service.
  4. If the Reason names an address in use, the netstat output above tells you which process holds the port; give one of the two a different fixed port under TCP/IP, IP Addresses, IPAll.
  5. If the log shows a 17190 line about the fallback certificate, or the instance has a certificate thumbprint configured, compare the configured value with the Cert Hash(sha1) line in the certutil output, character for character.

The order matters: read the Reason before changing anything. Every cause here produces the same three-error sequence in the log, so the numbers alone do not tell you which one you have.

If the log now announces the server listening on the port you expect, you are done. If not, the next section explains what happens in this phase of startup and which failure you are looking at.

Why it happens

Before it accepts a single login, SQL Server reads the protocol configuration written by Configuration Manager and binds a listener for every enabled protocol: shared memory, named pipes and TCP. If that cannot be done, startup aborts instead of continuing half-reachable, which is why the service appears to start and then stop a second later.

Error 17182 is that failure reported by the network interface layer. Its published text is “TDSSNIClient initialization failed with error 0x%lx, status code 0x%lx. Reason:” followed by a phrase in words. Two more entries follow it every time: 17826, published as could not start the network library because of an internal error in the network library, and 17120, published as SQL Server could not spawn a named thread, which in Microsoft’s own examples is FRunCM. Neither is worth chasing on its own.

Encryption is configured in the same phase, which is why certificates turn up in a networking error. SQL Server encrypts the login exchange whether or not you have asked for encrypted connections, so it needs a certificate at startup: the one you configured, or one it generates for itself. Error 17190 is published as “Initializing the FallBack certificate failed”, and it is specifically about that self-signed fallback rather than about a certificate you chose.

The hexadecimal values people quote from the 17182 line are Windows codes, not SQL Server ones. 0x2740 is Windows Sockets error 10048, WSAEADDRINUSE, published as address already in use, where typically only one usage of each protocol, address and port combination is permitted. 0x139F is Windows system error 5023, ERROR_INVALID_STATE, published as the group or resource not being in the correct state for the requested operation. Microsoft publishes no SQL-specific mapping for either, so read the Reason text rather than the number.

Every protocol is disabled for the instance

You have this one if The Reason on the 17182 line reads “All protocols are disabled”, the error is 0xd with status code 0x4, and Windows reports service-specific error 13.

  1. In SQL Server Configuration Manager, open SQL Server Network Configuration and then Protocols for your instance.
  2. Enable at least Shared Memory, which is what local connections need, and TCP/IP if anything connects over the network.
  3. Restart the service and confirm the log now announces the listeners.

The same values live in the registry under HKLM\SOFTWARE\Microsoft\Microsoft SQL Server\<instance>\MSSQLServer\SuperSocketNetLib, in the Enabled value of the Sm, Tcp and Np subkeys. Zero means disabled.

Something else already holds the port

You have this one if The Reason names the TCP listener and the failure started after another instance, an application or a proxy arrived on the same server.

  1. Identify the holder with netstat -ano | findstr :1433, then tasklist /fi "pid eq <pid>".
  2. If it is a second instance, give one of them a fixed and distinct port under TCP/IP, IP Addresses, IPAll.
  3. Restart the instance, then update connection strings, firewall rules and aliases to match.

Clients that connect by instance name rather than port also need the SQL Server Browser service running once you move off the default port.

The configured certificate thumbprint is not what you think it is

You have this one if The instance was configured for encryption by pasting a thumbprint into the registry, and it has not started since.

  1. Export the SuperSocketNetLib key for the instance to a .reg file and open it in Notepad using ANSI encoding. A question mark inside the thumbprint string is the documented signature of an invisible character copied out of the Certificates snap-in.
  2. Get the value cleanly with certutil -store my and take the string from the Cert Hash(sha1) line, with the spaces removed.
  3. Write the corrected value back and restart the service.

The fallback certificate cannot be created

You have this one if A 17190 line naming the FallBack certificate, usually after a service account change or a machine rebuild.

  1. Check that the service account can read the machine key store, since that is where the self-signed certificate is created.
  2. If a certificate is configured, grant the service account read access to its private key in certlm.msc under Personal, All Tasks, Manage Private Keys.
  3. Confirm the certificate has not expired and has a private key attached.

The instance starts before the network is ready

You have this one if The failure only ever happens at boot; starting the service by hand a minute later works perfectly.

  1. Set the SQL Server service to Automatic (Delayed Start) so it launches after the network stack has settled.
  2. Where the address belongs to a teamed, clustered or virtual adapter, prefer Listen All over binding to one address.
  3. Reboot once, deliberately, to confirm.

Full reference

Reading the 17182 line properly

The message is built from three pieces and people usually quote only the first. The error value and the status code are Windows numbers; the Reason is English and is the part Microsoft’s own troubleshooting articles key on. If you are pasting this error to somebody, paste the whole line and the two lines after it.

What you see Where it comes from
error 0xd, status code 0x4, Reason: All protocols are disabled Documented: every protocol is turned off for the instance
error 0xd, status code 0x38 or 0x1, with a certificate configured Documented: the thumbprint in SuperSocketNetLib is not a clean value
0x2740 anywhere on the line Windows Sockets 10048, address already in use
0x139F anywhere on the line Windows system error 5023, the resource is not in the correct state
A 17190 line after it The self-signed fallback certificate could not be initialised

Where the network configuration actually lives

Configuration Manager is a front end to one registry key per instance. Reading it directly is useful when the service will not start and you want to know what it is about to try.

Path under the instance’s MSSQLServer key Holds
SuperSocketNetLib\Sm\Enabled Shared memory on or off
SuperSocketNetLib\Tcp\Enabled TCP/IP on or off
SuperSocketNetLib\Np\Enabled Named pipes on or off
SuperSocketNetLib\Certificate The thumbprint of the configured certificate

Export the instance’s SuperSocketNetLib key before editing it. A mistyped value there stops the service from starting at all, and the only clue you get is the same 17182 you are trying to clear.

Getting in while nothing is listening

  1. Stop SQL Server Agent.
  2. From the instance’s BINN folder, run the engine as a console application with minimal configuration: sqlservr.exe -c -f -s INSTANCENAME.
  3. Connect locally over shared memory with sqlcmd -S lpc:.\INSTANCENAME -E.
  4. Make the change, stop the console instance with Ctrl+C, and start the service normally.

Things this error is not

  • It is not a firewall problem. A firewall blocks traffic to a listener that exists; 17182 happens before any listener is created.
  • It is not a licensing or edition problem. Nothing about binding an endpoint depends on which edition you run.
  • It is not a login problem. No login has been attempted at the point this fails.
  • It is not usually a cluster problem, even on a cluster. The protocol configuration, port reservations and certificate permissions are per machine, so compare the failing node with the working one rather than the cluster configuration.

After the instance is up again

If you changed the port, everything that connects by number needs updating: connection strings, aliases, firewall rules and any monitoring that checks the old port. If you cleared a certificate to get running, put a proper one back and grant its private key to the service account rather than leaving the instance on its self-signed fallback indefinitely.

Every code this article covers

Code What it points at Source
17182 TDSSNIClient initialisation failed, with an error value, a status code and a Reason in words on the same line Microsoft Learn
17826 Could not start the network library because of an internal error in the network library; logged after 17182 Microsoft Learn
17190 Initialising the FallBack certificate failed – the self-signed certificate SQL Server generates for itself, not a certificate you configured Microsoft Learn
0x2740 Windows Sockets error 10048, WSAEADDRINUSE: address already in use, since only one usage of each protocol, address and port combination is permitted Microsoft Learn
0x139F Windows system error 5023, ERROR_INVALID_STATE: the group or resource is not in the correct state to perform the requested operation. Microsoft publishes no SQL-specific meaning for it Microsoft Learn
17120 SQL Server could not spawn a named thread; Microsoft’s 17182 examples show it as FRunCM, logged once the network library has failed Microsoft Learn

Confirm the fix worked

  1. The current error log announces the server listening on the addresses and port you expect, with no 17182 after it.
  2. netstat -ano | findstr :1433 shows a LISTENING entry owned by the sqlservr process id.
  3. Test-NetConnection yourserver -Port 1433 succeeds from a client machine, not just from the server.
  4. An ordinary application connection string works, not only a local one.
  5. Restart the service once more and confirm it comes up unattended.

Questions people ask about this

Is this a firewall problem?

No. A firewall blocks traffic to a listener that already exists. Error 17182 happens before any listener is created, so no packet gets far enough for a firewall rule to matter. Firewall rules become relevant only once the log says the instance is listening.

Does fixing this cost anything?

No. Every documented cause is configuration: disabled protocols, a port conflict, a bad certificate thumbprint, a private key permission or startup ordering. No licence or edition affects whether the engine can bind its endpoints.

Should I disable encryption to get past 17190?

Clearing a broken certificate configuration is a reasonable emergency step, and SQL Server will still encrypt the login exchange using a certificate it generates itself. Put a proper certificate back afterwards and grant its private key to the service account.

Why does the instance run on one cluster node and not the other?

Because everything in this article is per machine. The failing node has different protocol settings, a different port conflict, or a certificate whose private key the local service account cannot read. Compare the two nodes’ network configuration rather than the cluster resource.

The log shows 17826 and 17120 but no 17182. What now?

Scroll up. Those two are logged as consequences of the network layer failing, and the 17182 line with its Reason is above them. If the log genuinely starts at 17826, you are looking at a truncated or recycled log; check the previous ERRORLOG file in the same folder.

Related error codes

Was this article helpful?

Your feedback helps us improve our documentation.

Related articles

License Error Error 3967: Insufficient Space in tempdb to Hold Row Versions Free Fix Error 102: Incorrect Syntax Near – Finding the Real Line in a Long Batch Free Fix Error 512: Subquery Returned More Than One Value, and Related Query Failures License Error Error 17051: The SQL Server Evaluation Period Has Expired – What To Do Now
โ† Back to Knowledge Base