Insert statement with random generated number - sql-server

Question

What I'm trying to accomplish is to assign a 5 digit number to a row in the table and insert that value with column value to a different table.
For example, this query returns all the rows that I would like to assign a 5 digit random ID:
SELECT DISTINCT(ExternalAgentId) FROM lOGS_V WHERE EXTERNALAGENTID <> ''
Currently it's returning 4600 rows. NOTE* ExternalAgentID is varchar(50)
What I need to do is assign a random 5 digit number between 10001 and 39999. Once I generate the number insert it to the table with ExternalAgentId from previous query to another table.
Here's the approach I took:
DECLARE #randAgentID int;
DECLARE #AgentID int;
DECLARE #MIN INT;
DECLARE #MAX INT;
SET #MIN = 10001
SET #MAX = 39999
SELECT #randAgentID = ROUND(((#MAX - #MIN -1) * RAND() + #MIN), 0)
SELECT #AgentID = InternalAgentID FROM VendorAgentIDs where InternalAgentID = #randAgentID
IF #AgentID is null
BEGIN
INSERT INTO VendorAgentIDs (VendorID, TRIAgentID) values (SELECT DISTINCT(ExternalAgentId) FROM LOGS_V WHERE EXTERNALAGENTID <> '', #randAgentID)
END
ELSE
BEGIN
SELECT #randAgentID = ROUND(((#MAX - #MIN -1) * RAND() + #MIN), 0)
INSERT INTO VendorAgentIDs (VendorID, TRIAgentID) values (SELECT DISTINCT(ExternalAgentId) FROM LOGS_V WHERE EXTERNALAGENTID <> '', #randAgentID)
END
It's generating random 5 digit number correctly. However, I'm having two issues:
I have a syntax error in my insert statements.
...values (SELECT DISTINCT(ExternalAgentId)...
If the randAgentID exists in the table, it goes to the ELSE statement. I know the chances are low, but what if the next randAgentID also exists in the table? How can I prevent that?
The table VendorAgentIDs has three columns.
ID (autoincrement)
VendorID (varchar(50))
InternalAgentID (PK, int)
Any suggestions on how I can resolve the above two issues?
Thanks
* *UPDATE
Based on the suggestion, I edited my code. However, I now have a violation of PRIMARY KEY.
DECLARE #randAgentID int;
DECLARE #AgentID int;
DECLARE #MIN INT;
DECLARE #MAX INT;
SET #MIN = 10001
SET #MAX = 39999
SET #AgentID = 1
WHILE #AgentID IS NOT NULL
BEGIN
SET #AgentID = NULL
SELECT #randAgentID = ROUND(((#MAX - #MIN -1) * RAND() + #MIN), 0)
INSERT INTO VendorAgentIDs (VendorAgentID, InternalAgentID) SELECT DISTINCT(ExternalAgentId), #randAgentID FROM LOGS_V WHERE EXTERNALAGENTID <> ''
SELECT #AgentID = InternalAgentID FROM VendorAgentIDs where InternalAgentID = #randAgentID
END
Violation of PRIMARY KEY constraint 'PK_VendorAgentIDs'. Cannot insert duplicate key in object 'dbo.VendorAgentIDs'.

Answer 1

May be this will solve your PK issue
INSERT INTO VendorAgentIDs
(VendorAgentID,InternalAgentID)
SELECT DISTINCT ExternalAgentId,#randAgentID
FROM LOGS_V
WHERE EXTERNALAGENTID <> ''
AND ExternalAgentId NOT IN (SELECT VendorAgentID
FROM VendorAgentIDs)

Answer 2

INSERT INTO VendorAgentIDs (VendorID, TRIAgentID) values (SELECT DISTINCT(ExternalAgentId) FROM TRI_PORTAL.DBO.SCREENPOPLOGS_V WHERE EXTERNALAGENTID <> '', #randAgentID)
Is invalid SQL. However I think you can do
INSERT INTO VendorAgentIDs (VendorID, TRIAgentID) SELECT DISTINCT(ExternalAgentId), #randAgentID FROM TRI_PORTAL.DBO.SCREENPOPLOGS_V WHERE EXTERNALAGENTID <> ''
As for the possiblity (eventuality) of a collision you should just do a loop. It becomes increasingly inefficient but if it has to be a random number and not an index then, so be it.
SET #AgentID = 1
WHILE #AgentID is not null
BEGIN
SET #AgentID = NULL
SELECT #randAgentID = ROUND(((#MAX - #MIN -1) * RAND() + #MIN), 0)
SELECT #randAgentID -- why is this here?
SELECT #AgentID = InternalAgentID FROM VendorAgentIDs where InternalAgentID = #randAgentID
END