Transactional Dead-letter messages

Question

Hi Folks,

I've noticed some major slow downs in our NPM environment lately and specifically around deleting nodes or interfaces. In my investigations I noticed that on a daily basis there are messages building up in the Transactional Dead-letter message queue on the NPM application server, the messages all to be from the information service. My understanding of this queue is limited and I'm unsure how to troubleshoot this.

I have had a look at the avg disk queue length on the DB server but all appears good there. Any advice there would be appreciated.

thanks

Ciarán

jposner · Answer

Would it be best to do these steps on the main and all additional pollers prior to running the database queries?

Ciag · Answer

Hi cmatrask,

if you want to clear/purge the messages you can easily do this by logging into your polling server right click the queues your messages are stuck on, in the all tasks menu you should be able to 'Purge'

My issue was that after I cleared them more appeared soon after. In the end I had to raise a ticket. For me the solution was the steps below;

Shut down all orion services 
 - Open c:\documents and settings\all users\application data\solarwinds\jobengine\Data
 - Make a copy of the JobEngine35 - Blank.sdf file.
 - Rename the JobEngine35.sdf file to JobEngine35.sdfold
 - Rename your copy of JobEngine35 - Blank.sdf to JobEngine35.sdf
 - Right click the new JobEngine35.sdf and open Properties.
 - Uncheck the read-only attribute.

Then run the following query against the Orion database, it would be best to create a backup copy of the Orion database before running the queries.

DELETE FROM PendingNotifications
 DELETE FROM SubscriptionTags
 DELETE FROM Subscriptions

Start up all of your Orion services again

HTH

Ciarán

cmatrask · Answer

I have the same issue. I would like to delete them all.