When a server fails, it stops announcing itself. When the master browser does not receive a server announcement for three of the server's current announcement periods, the master browser removes the non-browser from the browse list. It might take up to an additional 15 minutes for the backup browsers to retrieve the updated browse list from the master browser, so it could take as long as 51 minutes from the time a server fails to when it is removed from all browse lists.
Because a backup browser announces itself in the same way as a server, the procedure when a backup browser fails is the same as that for a server. If the name of this backup browser has been given to any clients, attempts made by those clients to contact this backup browser fail. The client then retries the NetServerEnum API call on another backup browser on the client's list of browsers. If all the backup browsers that a client knows have failed, the client attempts to get a new list of backup browsers from the master browser. If the client is unable to contact the master browser, it forces a browser election.
When a master browser fails, the backup browsers detect the failure within 15 minutes. After a master browser failure is detected, the first backup browser to detect the failure forces an election to select a new master browser. In addition, it is possible that between the time the master browser fails and the election of a new master browser happens, the domain will disappear from the list of domains in the browse list. If a client performs its first NetServerEnum API call after the old master browser has failed but before a backup browser detects the failure, the client forces an election. If a master browser fails and there are no backup browsers, browsing in the workgroup or domain will not function correctly.
When a domain master browser fails, other master browsers see only servers on the same local subnetwork. Eventually, all servers that are not on the local subnetwork are removed from the browse list.