Alerting

• jblovesthegym

  Ten Centuries

  Points: 1048

  More actions

  April 17, 2012 at 4:57 am

  #260151

  Hi all,

  We have an ongoing situation where our DBA team is being overalerted, with even trivial issues (like a backup succeeding) being flagged as 'Critical', and as such they are being overwhelmed with alerts, in addition to existing workload.

  Over the next week or two I want to investigate what software is generating these alerts, who gets them etc, and if anyone has a list of best practices, I'd like to have it.

  We have Idera DM in operation on one of our clusters, but at the moment I'm only using its dashboard, which I refer to as part of my morning checks, and then several times a day - not the ideal method of keeping on top of things.

  What I envisage for alerting categories would be the following:

  Critical - issue that merit an immediate response, ie a backup job failing, service stopped etc;

  Moderate - An issue that should be looked at immediately and a response scheduled when convenient to the business, ie backup jobs running over x seconds, indexes fragmented over x% etc

  Informational - Trivial issues, ie a known user filling a disk to just over 80% for a few seconds during staging ops.

  I would like some input on what kind of alerting you good people do, best practices, even if it's for versions other than 2K.

  Cheers,

  JB

• VastSQL

  SSC-Dedicated

  Points: 36188

  More actions

  April 17, 2012 at 7:52 am

  #1474728

  Check the email from address that will give you a good idea from where its generation if its giving alert through email.

• jblovesthegym

  Ten Centuries

  Points: 1048

  More actions

  April 17, 2012 at 8:09 am

  #1474744

  Ratheesh.K.Nair (4/17/2012)

  ---

  Check the email from address that will give you a good idea from where its generation if its giving alert through email.

  I had already set up a meeting to look at those mails, but what I asked for above was an idea of current best practice, ie I check to see if any scheduled jobs have failed. I want to move to a point where I am not doing any manual checks of anything except to see if my own monitorin software is running!