Modifying Alert Settings

Is there a method to adjust the alert settings. I find some of the setting too restrictive (such as level 3 warning at 80% volume capacity). Addtionally, the email alert for instance unavailable is way too sensitive for a very large network. We would desire the ability to generate the alert after a certain level of communication loss. I am current getting more than 4000 email alerts a day.

  • While we don’t have an interface for modifying alert settings, I’m checking with the team to see if this can be modified through a configuration change. If I may ask, what version are you using? We adjusted the sensitivity for the instance unavailable alert in 1.1 but if it remains too sensitive we may need to take another look at it. 4000 emails per day is clearly excessive!

  • Currently using v1.1. We may be a special case as our network is so distributed (end-to-end network more than 1000 geographically)
    Possibly your team might want to consider using the 600ms rule. This generally will prevent false alerts due to slow network routes.
    Ultimately will be monitoring more than 1200 instances. Thank you!

  • I’d like to bump this topic as we have similar questions that are posted by Vicky in regards to be able to modify the thresholds by some setting. We also use the heavily configurable SQL Diagnostic manager on a few servers, and love the capability to groom the thresholds individually.
    With Elements we have now a bunch of Alerts for example the “volumes at risk of filling up”. We have some volumes on which we have fixed size files that will not grow and would for example like to exclude them from Alerting.
    We have not gotten yet the ones Steven describes, but potentially could get a few hundreds of those as well.

    Are there any plans to allow customization of Alert thresholds, preferably at instance level?

  • Hey folks. Sorry to be such a late comer to this discussion. There ARE plans to provide more configurability for collection, retention, and alerting. I don’t expect it to ever be quite to the level that diagnostic manager provides however. In fact, recently, the product focus has shifted from a health monitoring tool to more of an inventory tool. On that note, it is about to re-release as “SQL Inventory Manager”. We’ve added functionality useful to an inventory tool like custom fields for the instance, tagging of databases, some extra options around discovery, and a few others. There are also some stability fixes and performance enhancements, and integration into our new “Dashboard”, where the other tools are headed, and Diagnostic Manager and Business Intelligence manager are today.

    Not to completely ignore your question though. Here is a script you can run that will show you all of the thresholds for alerting in SQL Elements. You can then go adjust the appropriate values in the alert_thresholds table to suit your needs.


    use IderaElementsRepository
    SELECT alert_metadata.metric_id, alert_thresholds.metric_id, alert_metadata.name, alert_thresholds.critical, alert_thresholds.warning, alert_thresholds.informational, alert_metadata.description
    FROM alert_metadata INNER JOIN
    alert_thresholds ON alert_metadata.metric_id = alert_thresholds.metric_id

    After you have adjusted thresholds, you can also adjust the timing of the collection of various sets of information. Someone asked us today, could they monitor the space used on their drives at a tighter interval. By default we do that every hour. Here is what that looks like.

    SELECT i.[name] as Instance, s.[instance_id], s.[job_type]
    ,s.[job_description]
    ,s.[active]
    ,s.[run_every]
    ,s.[days_of_week]
    FROM [IderaElementsRepository].[dbo].[job_schedules] AS s
    LEFT JOIN [IderaElementsRepository].[dbo].[instances] AS i ON
    i.[instance_id] = s.[instance_id]
    ORDER BY instance_id, job_type

    the run_every column tells SQL Elements how often to run that collection. This number is in seconds. The first few lines not associated to an instance are for discovery, the rest are assigned to each instance to pull the corresponding data.

    So, long story short, at present, if you want to adjust the alerting thresholds or collection intervals it is a bit of a manual task.