Sunday, June 4, 2017

SQL SERVER DBA “INTERVIEW QUESTIONS AND ANSWERS”



Following are some “sql server dba interview questions” to help you in interviews.
I have got many requests to publish first set of MSSQL DBA Interview questions. I observed that many interviewers are interested in finding out whether you worked on DB corruption or not. Here you go http://mssqlcorruptiontackle.blogspot.com/2010/12/ms-sql-corruption.html . Go ahead with ten most important Microsoft SQL Server DBA Interview Questions.
Interviewers are testing at least basic knowledge of clustering. Read following wires to be ready for this surprise: MSSQL Interview Question Clustering- Part1Part2Part3 and Part4


  1. What is the difference between lock, block and deadlock?
  2. What is the meaning of lock escalation and why/how to stop this?
  3. How to truncate the log in sql server 2008?
  4. How to Start Service Using Powershell Commands?
  5. What changes in the front end code is needed if mirroring is implemented for the high availability?
  6. Where does the copy job runs in the logshipping… Primary or secondary?
  7. What are the ways to find what code is running for any spid?
  8. When you get following error? Error 3154: The backup set holds a backup of a database other than the existing database.
  9. Does DBCC CHECKDB requires DB to be in SINGLE_USER mode?
  10. How to view the error log for any specific instance?


Want to prepare in ten minutes - read article Interview and Beyond in 10 Min
———————————————-
Answer 1:
Lock: DB engine locks the rows/page/table to access the data which is worked upon according to the query.
Block: When one process blocks the resources of another process then blocking happens. Blocking can be identified by using
SELECT * FROM sys.dm_exec_requests where blocked <> 0
SELECT * FROM master..sysprocesses where blocked <> 0
Deadlock: When something happens as follows: Error 1205 is reported by SQL Server for deadlock.
Answer 2: When the DB engine would try to lock page first and then it escalates locks to page and then table. If we
understand that whole table would be locked for the processing thenn this is better to use TABLOCK hint and get complete table blocked. This is a nice way to avoid the wastage of sql server DB engine processing for lock escalation. Somewhere you may also need to use TABLOCKX when you want an exclusive lock on the table in the query.
Answer 3: BACKUP LOG TestDB WITH TRUNCATE_ONLY is gone. SQL server doesn’t allow you to truncate the log now otherwise whole purpose of a DB is defeated. Read article http://mssqlcorruptiontackle.blogspot.com/2010/12/mssql-server-dbcc-loginfo-status-2-log.html to surprise interviewer with your answer. You have to make sure whether you need log or not. If you don’t need log then have the recovery model simple instead of full. If you don’t want the log to be accumulated in some particular bulk logging then change the recovery model BULK LOGGED for that duration and take one tlog backup just before and after this change. I shall discuss this later in my later blog. BACKUP LOG command backs up the t-log and frees the space in the log file.
Answer 5: You need to add only FAILOVER PARTNER information in your front end code. “Data Source=ServerA;Failover Partner=ServerB;Initial Catalog=AdventureWorks;Integrated Security=True;”.
Answer 6: Secondary server. This question is basically asked to find out whether you have a handson work on logshipping or not. I came through many cases when candidates have mentioned logshipping experience in many projects and they can’t answer this question. I never selected any candidate if he/she don’t answer these kind of small questions.
Answer 7: Well there are many ways to do this.
1. find the spid which you want to analyze. An spid is assigned as soon as a client connection is established with the SQL server. To find the spid you can run any of the following command:
a) SP_WHO2 ‘ACTIVE’ — This will give you only active spids.
b) SELECT * FROM sys.dm_exec_requests
2. Get the spid from above two queries and use any of the following query to get what is happening behind that spid.
a) dbcc inputbuffer(<spid>)
b) sql2005 and sql2008 – SELECT * FROM sys.dm_exec_sql_text(<sqlhandle of that spid>)
c) sql2005 and sql2008 – SELECT * FROM fn_get_sql(<sqlhandle of that spid>)
Answer 8: The error comes when you are trying to restore the DB which already exists. Use WITH REPLACE option to restore the DB with a different name.
Answer 9: Yes and No. This is tricky question. If you are using repair option with CHECKDB then you have to have the DB in single user mode. Following is the method to have your DB in a single user mode.
Use master
go
sp_dboption dbname, single, true
Following is the error which you get when you run the DBCC CHECKDB with repair option w\o having the DB in single user mode.
The same is true for DBCC CHECKDB also.


Answer 10: There are many ways but I prefer following method. Take a scenario when you want to find the error log when the DB was put in a single user mode.
CREATE TABLE #Errorlog (Logdate Datetime, Processinfo VARCHAR(20),Text VARCHAR(2000))
INSERT INTO #Errorlog
EXEC xp_readerrorlog
SELECT * FROM #Errorlog
WHERE Text Like ‘%SINGLE%USER%’
Most of the readers of this blog also read following interview questions so linking them here:
Powershell Interview Questions and Answers
SQL Azure Interview Questions and Answers Part – 1
SQL Azure Interview Questions and Answers Part – 2
Read following blogs to surprise your interviewer and create chances to get into IT sector:
  1. https://tuitionaffordable.wordpress.com/sql-server-dba-interview-questions/
  2. http://mssqlcorruptiontackle.blogspot.com/2010/12/ms-sql-corruption.html
  3. http://mssqlcorruptiontackle.blogspot.com/2010/12/sql-server-2008-r2-foreign-key-delete.html
  4. http://mssqlcorruptiontackle.blogspot.com/2010/12/mssql-server-dbcc-loginfo-status-2-log.html
  5. http://mssqlcorruptiontackle.blogspot.com/2010/12/ssrs-report-adding-header-and-footer.html

No comments:

Post a Comment