Why a Fragmentation Alert with NONE over %?

SQL Server performance monitoring and alerting

Moderators: eddie davis, priyasinha, Adam, chriskelly, Chris Lambrou, Chris Spencer

Why a Fragmentation Alert with NONE over %?

Postby PDinCA » Mon Apr 29, 2013 6:54 pm

Version 3.3.0.270 issued an Alert for Fragmentation via email but there is no Alert in the UI. This appears to be a two-part bug... 1) No Alert listed and 2) No Indexes actually qualify - see bolded-red below.

Fragmented indexes | 2687013


Raised on: <Servername> (local) > xXXXXXX
Time raised: 24 Mar 2013 1:00 AM (UTC-04)
Time ended: 28 Apr 2013 1:00 AM (UTC-04)
Details
Database name: xXXXXX

Total number of indexes: 106
Indexes above fragmentation threshold: 0

Indexes last checked: 28 Apr 2013 1:00 AM
Fragmented indexes:
Index Table name Fragmentation(%) No. of pages
PDinCA
 
Posts: 516
Joined: Mon Jul 25, 2005 11:42 pm
Location: Costa Mesa, CA, USA

Postby chriskelly » Wed May 01, 2013 7:16 pm

This looks like the content of an Alert Ended alert email, which is confirmed by the "Time ended" detail.

If we work to this assumption, then this explains why it contains the entry "Indexes above fragmentation threshold: 0" as this would end the conditions that led to the alert being generated (would lead to the alert being marked as 'ended'). Given that the alert was generated over a month ago, and assuming that the purge settings are still set to default, then once the alert 'ended' then the data would have been purged.

I hope this makes sense.
Chris Kelly
Technical Support Engineer
chriskelly
 
Posts: 330
Joined: Mon Apr 19, 2010 1:44 pm
Location: Cambridge, UK

Postby PDinCA » Wed May 01, 2013 8:02 pm

Makes sense, especially as I have a 2-week retention period.

Would you consider:
1. NOT purging "Open" Alerts like this.
2. Including the ORIGINAL Alert reference in the "Ended" email so it can be shown in the UI with one click.
3. Including a BIG BOLD "ALERT SPANS <LONG DURATION>" chunk of verbiage in the email to bring our attention to this kind of thing - I seriously don't have time to read every alert in detail so the more you can highlight "Weird stuff" the more help it is.

<LONG DURATION> should be anything more than the data purge period and optionally, configurable, perhaps, anything over X days. I typically only care about the ORIGINAL Alert email and not the Ended one for 95%+ of reported incidents - I investigate the original, deal with it, clear it, and move on. To be "woken up" so to speak, with really OLD NEWS is noise I don't need so anything you can do to reduce the noise level of Ended emails would be brilliant!

Cheers!
PDinCA
 
Posts: 516
Joined: Mon Jul 25, 2005 11:42 pm
Location: Costa Mesa, CA, USA

Postby chriskelly » Wed May 01, 2013 8:35 pm

1. From the information available it looks like the alert was purged after it Ended, despite it exceeding the 2 week purge window.
2. The URL would lead to the Alert, if it was still available. Notice in the URL is the unique id of the alert, this does not change and always leads to the same alert details page. There is not a different one for different alert statuses.
3. This could be considered for a future release.

Finally, you can configure the email settings so that it doesn't send out email 'Ended' status emails. This can be configured globally, for groups, individually etc.
Chris Kelly
Technical Support Engineer
chriskelly
 
Posts: 330
Joined: Mon Apr 19, 2010 1:44 pm
Location: Cambridge, UK

Postby PDinCA » Wed May 01, 2013 9:27 pm

Thanks, Chris, just goes to show how little time one has to read the user guide, despite having been an original user of SQL Response, then SQL Monitor!

Thanks for the info.
PDinCA
 
Posts: 516
Joined: Mon Jul 25, 2005 11:42 pm
Location: Costa Mesa, CA, USA


Return to SQL Monitor 3

Who is online

Users browsing this forum: No registered users and 0 guests