As per Microsoft: "The BUILTIN\Administrators login does not have access to SQL Server or the BUILTIN\Administrators login has been removed from SQL Server, and the SQLAgent service is starting under the LocalSystem account.
The issue does not occur if the SQLAgent service is starting under a domain or computer account that is listed separately in SQL Server Enterprise Manager and that has System Administrators (sysadmin) privileges". See ME237604
for a workaround.
From a newsgroup post: "It appears that the SQL Server service fails to start on the other node. SQL Server Agent depends on SQL Server Service. Since SQL Server is failing, SQL Server Agent will fail too. To find why SQL Server failed, please review the SQL Server Error logs. The reason, the group fails over to the other node is that by the cluster service will try to start the SQL server service 3 times. If it fails 3 times in the default time (900s), the group is failed to other node. If you do not want the failure of a resource to affect the group, you can right click the resource and select properties, click Advanced and uncheck "Affect the group". Also, to find the actual service that is failing do the following:
1) Right click the SQL group and bring it offline.
2) Move the group to other node (where resources fail).
3) Now start bringing resources online (one at a time). Start with disk/s, followed by IP, followed by network name then SQL Server, followed by SQL Server Agent.
From the errors, you will find that the SQL Server will fail to come online.
Also, for this test, uncheck "affect the group" for SQL Server resource. If you don’t do this, then it will try to come online 3 times and if it fails all 3 times, it will affect the group and fail the whole group to previous node. Once you know that SQL Server service failed to come online. Review the SQL Server error logs to get detailed information. If SQL Server service comes online, then start SQL Server Agent resource. If it fails, review the SQL Server Agent logs (located in the LOG folder - same place where the SQL error logs are located)".
From a newsgroup post: "If you also receive event id 17055 with the following description: "The maximum limit for connections has been reached", then you should try the following: Start the SQL server in single user mode. To do this, go to control panel, services. Stop the server and before you restart it, type “-m” in the startup parameters and see if you can get in that way.
See if you can check the connections property of the server. It if has been set to a static number, try changing that to a dynamic setting".