Submit a ticketCall us

Looking to compare latest NPM features with previous versions of NPM?
The NPM new feature summary offers a comparison of new features and improvements offered with this release.

 

Home > Success Center > Database Performance Analyzer (DPA) > The proper care and feeding of DPA monitors

The proper care and feeding of DPA monitors

Table of contents
No headers
Created by Interspire Import, last modified by MindTouch on Jun 23, 2016

Views: 14 Votes: 0 Revisions: 8

Some of the most frequently asked questions from our customers are:

  • How do I know if Ignite is running properly on my monitored database?
  • Or, why is my busy production database showing ‘Idle or lightly used’ in Ignite?

These types of questions are often centered around the operation and performance of Ignite’s QuickPoll Monitor processing. In order to answer these questions, its neccessary to understand how Ignite’s monitor processing really works. Basically, there are 3 types of monitoring that QuickPoll performs: the QuickPoll (or Performance Gathering) Interval, the SummaryPoll Interval and the SQL Text Interval.

How QuickPoll Monitoring Works

The QuickPoll monitor connects to a monitored database to bring back raw performance data to the repository database for further analysis and interpretation at a predefined interval of time (the default interval is every second). The SummaryPoll process then interprets/compiles this information in the repository database so that it can be displayed in the Ignite GUI. The SummaryPoll’s default interval is set to run every 10 minutes. Finally, the SQL Text Interval is a predefined setting which tells the QuickPoll Monitor when to retrieve SQL Text information from the monitored database. Since SQL Text is often kept for periods of times in the database’s cache, we do not need to get that information as frequently. The default setting for retrieving SQL Text is set to every 2 minutes. You can review the interval settings for your monitored databases by choosing the “File > Database Options” menus in Ignite. An example of the 3 settings are listed in Figure 1. below:


blog1.jpg
Figure 1 – Quck Poll, Summary Poll and SQL Text Poll Settings

In addition to the Interval settings in the Database Options menu, you can get a quick idea of how well your Quickpoll monitoring is performing by clicking on the blue colored bar next to the database name (see red arrow). In the above example, (Figure 1), the QuickPoll monitor is performing at 95% capactiy. Now you might ask, ‘What does that mean exactly’? As I stated earlier, the QuickPoll process brings back raw data from the monitored database at a predefined interval. In other words, it samples data at the predetermined interval’s rate. In our example, the interval for QuickPoll is set to ’1 poll per second’and our SummaryPoll interval is set to run ‘every 10 minutes’. So,if QuickPoll runs at 100%, it should be retrieving 600 samples of data in the 10 minute timeframe prior to SummaryPoll’s processing. That is, 1 poll per second * 60 seconds * 10 minutes = 600 polls. So, if our QuickPoll monitoring is running at 95%, we are getting 570 samples of raw data per Summarization (600 polls * 95% = 570).

The accuracy of Ignite’s performance intelligence statisitics is dependant on how well QuickPoll is gathering performance data from the monitored database. In general, the closer to 100% in collecting data the better. A good rule of thumb when looking at QuickPoll’s performance is to make sure that it’s running at 80% capacity or above.

Trouble Shooting QuickPoll’s Performance

If you find that your QuickPoll monitoring is running less then 80% and/or you are frequently getting ‘Idle or lightly used’ messages in Ignite, there are several items you can check before contacting Confio Support. I’ve listed the check list below:

  • Check Ignite’s Error Log for QuickPoll connectivity errors to the monitored database. If the QuickPoll process is having trouble connecting or keeping connected to the monitored database, it probably isn’t getting a chance to gather much information. Resolve the connectivity issue from the Repository to the Montiored database.
  • Check Ignite’s Error Log for any repository errors such as space issues, or any ORA- errors and resolve these issues.
  • Check Ignite’s Error Log for any SQL Text warning messages such as ‘Text Poll Interrupted because duration exceeded 120 seconds (x of Y SQLs were remaining)’. If you are seeing these errors frequently, further investigation is required (Refer to the next step below).
  • Use the Repository Query Tool in Ignite (Tools menu), to look at QuickPoll’s performance. There are several stored SQLs which can be used to help you evaluate QuickPoll’s performance. In Figure 2 below, the first step is to find the ID of the monitored database which is having the monitoring issues. In this example, the CUSTREPO_LAB1 database has the ID of 4.


jg_blog2.jpg
Figure 2 – Find ID for Database Having Monitoring Issues

In the Stored SQL drop down list, you can choose and modify queries that give you more information on QuickPoll, SummaryPoll and SQLText Poll performance. For convenience, I’ve listed an example of a query here:

select activity, min(execs), max(execs), min(totalrows), max(totalrows), max(totaltime) total_time, max(avgtime) average_time
from contime
where d > sysdate-1
and (activity = ‘QUICKPOLL’ or activity like ‘SP-%’ or activity like ‘%SQLTEXT%’)
and dbid = 4
group by activity

This query will show the maximum total times and average times for the QuickPoll, SummaryPoll and SQL Text processes for a specific database. Generally, high total_times and average_times could indicate a problem with your monitoring. For example, if the average time is high for the QuickPoll monitor, Ignite may be having connections issues to the monitored database or the monitored database might be too busy to service QuickPoll’s request for data. Or, if the SummaryPoll Total time is consistently high, the repository database may need further tuning in order for summarization to quickly complete. Finally, if the SQL Text Total time is high, the QuickPoll monitor may be having trouble retrieving SQL Text from the monitored database.

My Monitoring Is Slow or Hanging

If you suspect Ignite monitoring is having issues and you’ve found that the total times or average times of monitoring are very high, you should open a technical support ticket at http://www.solarwinds.com/support/ticket/. Our Support team can review the data and assist you with configuration changes that will improve Ignite’s Monitoring performance.

Last modified
19:18, 22 Jun 2016

Tags

Classifications

Public