Loading

After upgrading Tableau Server and performing a server cutover, some internal clients still reach the old Tableau Server while others reach the new one.

Udgivelsesdato: Jun 12, 2026
Beskrivelse

After upgrading Tableau Server and cutting over to a new server (new machine/IP), some clients reach the new server correctly while others continue to load the old server.

This is not caused by a problem with Tableau Server itself, but is rather a name resolution (DNS) issue: the affected clients are resolving the Tableau hostname to the old server's IP address. Because the behavior splits by machine rather than by user, the cause is something machine-wide on the affected hosts or, more commonly, the DNS server those hosts resolve against, rather than anything inside Tableau.

Løsning

The fix is a customer-side networking change. Updating internal DNS records is the responsibility of the customer's network/AD team and falls outside the scope of Tableau Technical Support, but the steps below can be provided as general guidance to help identify and resolve it. Work through them in order.

 

Step 1 — Confirm where the wrong answer is coming from

This isolates whether the problem is local to the affected host or coming from DNS, and tells you which fix applies.

On an affected host and on a working client, open a command prompt and run:

  1. nslookup <tableau-hostname> on both machines and compare the IP addresses returned.
  2. If the affected host returns the old server's IP while the working client returns the new server's IP, the wrong answer is coming from DNS, because nslookup queries DNS directly and bypasses both the local hosts file and the client DNS cache. This rules out a local hosts file entry and a stale client-side cache as the cause.
  3. Note the Server: line at the top of each nslookup result. If the affected host and the working client are answered by different DNS servers, this confirms they resolve against different resolvers (split-horizon DNS), which is why only some clients were affected.

Step 2 — Rule out a local hosts file entry (only if Step 1 showed nslookup itself is correct)

A hosts file entry overrides DNS and is local to one machine. This is only the cause if nslookup returns the new (correct) IP but the host still reaches the old server.

  1. On the affected host, open C:\Windows\System32\drivers\etc\hosts in Notepad as an administrator.
  2. Look for any line mapping the Tableau hostname to the old server's IP. Remove or comment it out, then save.
  3. Run ipconfig /flushdns and retry.

Step 3 — Update the internal DNS record (the fix when Step 1 shows DNS is returning the old IP)

  1. The internal DNS A record for the Tableau hostname still points at the old server's IP and was never updated during the cutover. The customer's network/AD team will need to update it.

Step 4 — Flush caches and verify

After the record is updated, on an affected host:

  1. ipconfig /flushdns
  2. nslookup <tableau-hostname> — should now return the new server's IP.
  3. Browse to Tableau from the affected host. It should now land on the new server exactly as the working clients do.
Vidensartikelnummer

005386905

 
Indlæser
Salesforce Help | Article