SQL Phone Home: Teaching Your SQL Servers to Call for Help
1 like2,231 views
This document discusses using SQL Server Agent to automate monitoring of a SQL Server instance and alerting when problems are detected. It provides examples of things that could be monitored like server restarts, databases not being backed up, failed jobs, and more. The document advocates the principle of "see something, say something" by having SQL Agent queries check for issues and send emails or log events to call for help if problems are found. It also cautions that any advice found online should be tested before implementing.
SQL Phone Home: Teaching Your SQL Servers to Call for Help
- 1. SQL Phone Home Teaching SQL Server to Call for Help
- 4. Mike Hillwig SQL Server DBA Working with SQL Server since SQL 7 Senior DBA at hosting division of a financial software company owned by a large financial institution Resume includes Acme Packet Oracle, Shawmut Design and Construction, Equitable Resources Certified in ITIL Foundations
- 5. Beware of the Blogs There is some amazing advice out there. But… Anybody can put bad advice on the internet Trust people you know Assume I’m an idiot I don’t trust people who say “ALWAYS” or “NEVER” Test everything in your own test environment first.
- 6. Housekeeping Don’t let me talk too fast. Blog post at mikehillwig.com/go/phonehome Please complete your session evaluations.
- 7. What is this SQL Agent thing? SQL Server’s tool for automating processes Allows you to schedule things to run SQL Scripts Windows executables WMI scripts Can even watch for certain errors or situations and take action
- 8. What are we doing with it? Simple cause and effect If you see something, say something Run a query that looks for problems If problem is found, call for help
- 9. Spam vs Mail Today’s examples send mail In a perfect world, you’d log a ticket with your support system But for the sake of simplicity in our scripts, we’re sending mail today Make sure your SQL Agent can send mail Enable Mail XPs Also going to write to the Windows Application Event Log
- 10. My rule All alerts must be actionable. Everything else is noise.
- 11. What are some examples? 1. Server restarted 2. Autoshrink enabled. 3. Databases never backed up or no full backup in X hours 4. No transaction log backup in X hours 5. Default File Growths 6. Databases with datafiles that don’t grow 7. Failed jobs in the past 24 hours
- 12. See Something. Say Something.
- 14. Enough Talk Lets demo!
- 15. But wait. There’s more!
- 16. SQL Agent Alerting Several options Error message Severity condition Performance Conditions
Editor's Notes
- #3: SQL Server MVP Live in Boston Organizer of SQL Saturday and SQL Saturday BI Edition Pag dad Frequent flyer (this photo was taken in Amsterdam on my way to India in 2013) Native Yinzer/Pittsburgher
- #7: Good advice out there, but… Lots of outdated information. Do DB servers really need 2x RAM for PF?