Category: SQL Agent

Kenneth Fisher has some good advice when dealing with event notifications:

One of the most common ways to get an event notification is by email. So what happens when you get 500 emails in a day and only one or two are actionable? Do you read every single email? Spending quite literally hours to find those one or two gems? Or do you just ignore the whole lot and wait for some other notification that there is a problem. Say, by a user calling you?

Next, let’s say you have a job that runs every few minutes checking if an instance is down. When that instance goes down you get an immediate email. Which is awesome! Of course then while you are trying to fix the issue you get dozens more emails about the same outage. That is at best distracting and at worst makes it take longer for you to fix the issue.

Fun story time:  at one point during my work career, there was a person (not me!) who accidentally broke every single SQL Agent job on dozens of instances and nobody noticed it for hours.  These weren’t production instances so it wasn’t the end of the world or anything…except that included in the broken jobs were a bunch which ran every minute.  And alerted every minute.  Via e-mail.  The entire database team essentially lost e-mail access for 3 days as there were so many messages coming in that it overwhelmed our provider’s ability to serve messages to us.  This type of mistake can happen, and if we had put into place some of the things Kenneth talks about, the consequences would have been less severe.

Chris Bell has a way of making SQL Agent error messages a lot better:

Not very helpful. Sure, I know the job failed, and what step it failed on, but now I have to connect to the agent and look up the history to determine if this is something I have to worry about.

It would be nice to receive the details seen in the history of the job showing up in the email alert received.

Recently I have been working with systems that had all the alert on failure configured so we knew when things failed and could jump on re-running them if needed. We even had them showing up into a data team slack channel, so we had a history as well as notification to everyone on the team at the same time. The problem is that there were not any details in the alerts we received so we had to be able to connect and figure out what to do next or hope that our paid monitoring service would act on something after reading the details of the failure.

Chris has provided a script and gives some recommendations on job configuration which might reduce the number of alerts you get.

Chris Sommer has a gripe about the schedule_uid column in SQL Agent jobs:

When you script out a SQL Agent Job you’ll notice that the job schedule will have a schedule_uid parameter (providing your job has a schedule). The gottcha lies in that schedule_uid. If you create another job schedule with the same schedule_uid, it will overwrite the schedule for any jobs that are using it. i.e. Any other jobs that are using that schedule_uid will start using the new schedule. Normally I consider UID’s as very unique and chances of a collision are low, but if you do a fair amount of copying jobs between SQL Servers there’s a good chance this will bite you eventually. That’s what happened to us (more than once).

Lets see if I can explain it better in an example.

Something to think about when scripting SQL Agent jobs.

Lori Brown has a script to find jobs running longer than their 30-day average:

I needed to update some of our long running job monitoring code to improve it from the version that we have right now. I like this version because it uses msdb.dbo.syssessions (https://msdn.microsoft.com/en-us/library/ms175016.aspx) to validate that a job is actually running. I also wanted to know the percent difference between the current run duration versus an average duration per job from the past 30 days. I decided to place the calculated average into a table variable and then join on it to get my results. I also used the IIF function (https://msdn.microsoft.com/en-us/library/hh213574.aspx) to help me avoid a divide by zero error that comes up when the average duration equals 0.

One thing which could cut down on false positives would be to calculate the standard deviation as well.  I wouldn’t automatically assume that job executions were normally distributed, but if you look at things more than one standard deviation away from the mean, it should remove noise of jobs which are just a little over the average but not in dangerous territory.