This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.

You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

NPM, NCM server and pollers and the importance of versions

dragoon231 over 7 years ago

Hello everyone,

I have a server that runs NPM and NCM, with 2 additional pollers. There have been issues lately and I'm thinking it is because they run a newer version of NPM and NCM on the server, but then older versions on their 2 pollers. I could not find much information about this, I think everyone would expect it all to be upgraded at the same time. The issues range from webpage timing out, to just running slow, to gaps of information missing on the node page. The team restarts them and they are good to go for a bit (in regards to running). All 3 run the same version of windows (2008r2), etc. They all write to the same data base (sql 2008r2). I'm thinking this is the core of the problem.

I don't know how long they have been having the issues, nor do I know if an event coincides with when the problems started. My guess is , the main server was upgraded, not the pollers. In short this is causing issues with how they write to the database ( for example; how each version may define the column for something like CPU usage). I think this may be the biggest issue overall. Another issue I can think of by having a server with version A, and 2 pollers with version B could be the MIB and UnDPs?

Here are the versions:

Main server - NPM 11.5.2 , NCM 7.4

Poller 1 - NPM 11.0.4 , NCM 7.3.11

Poller 2 - NPM - 11.0.4, NCM- 7.3.11

Overall, my question is, is this a problem, or is there an acceptable level of compatibility between NPM versions?

Top Replies

0 neomatrix1217 over 7 years ago

All additional pollers should be on the same version as your main server please download the 11.5.2 additional poller package from your customer portal.
SolarWinds Online Help
Note: Any and all installed Additional Polling Engines and Additional Web Servers must use the same version that is installed on the primary Orion server.
Cancel
Vote Up +2 Vote Down

Cancel
0 Diner over 7 years ago

Exactly. The main reasons for matching the versions is usually potential difference in the database schema accross different versions both platform and modules. Other reason is inter-server communication in both directions. Example. pollers may notify the main poller about changes in polled data and the main poller must notify the poller about changes in "what to poll". Incompatible communication endpoints may potential cause waiting for response timeouts and following connection pool exhaustion or undeliverable messages cumulation. Consequences may be as you describe, thus the versions matching is pretty important.
Cancel
Vote Up +1 Vote Down

Cancel
0 dragoon231 over 7 years ago in reply to Diner

These are the answers I was looking for. The environment that is where the problem exists, I do not work on and do not have admin rights to even see settings in the web interface. I'm in the data center and manage SAM and NPM. I manage our environment, update, etc. but I have no pollers set up as the other team does for their NPM/NCM server. I had recommended well before posting this they upgrade the pollers, I just didn't know the specifics to tell them in regard to the consequences. Thank you for the responses everyone!
Cancel
Vote Up 0 Vote Down

Cancel
0 dragoon231 over 7 years ago

I currently have a ticket that was opened not long after I originally posted this. It was escalated to an engineer. For any curious I will update this when it is all complete. Since the post, I updated the pollers, ran the configurations and performed some database maintenance. You are able to access the web console and view settings and things like that, but if you click on a specific node, regardless of polling through an additional poller or the primary, the web console will time out. When trying to navigate using Hubble, it looks about like this:
Cancel
Vote Up 0 Vote Down

Cancel
0 neomatrix1217 over 7 years ago in reply to dragoon231

I would follow the advise on this KB article about clearing .Net
https://support.solarwinds.com/Success_Center/Network_Performance_Monitor_(NPM)/Slow_website_performance_with_a_high_per…
Cancel
Vote Up +2 Vote Down

Cancel
0 neomatrix1217 over 7 years ago in reply to dragoon231

https://support.solarwinds.com/Success_Center/Network_Performance_Monitor_(NPM)/Slow_website_performance_with_a_high_percentage_of_'Other_Time'_in_hubble
Cancel
Vote Up +2 Vote Down

Cancel
0 dragoon231 over 7 years ago in reply to neomatrix1217

Going through steps now. This looks familiar with the C:\Windows\Microsoft.NET\Framework\v4.0.30319\Temporary ASP.NET Files\root, this part i think the tech tried to do, but wouldn't let him because the www publishing service was still turned on. will let you know what happens
Cancel
Vote Up 0 Vote Down

Cancel
0 janene16 over 7 years ago in reply to dragoon231

dragoon231,
Since its the Other time that has the most time consumed, I would think there's a resource that is doing that, if not a dll file for the Node Details page being blocked. Can you answer these questions please?
1. When you customize the Node Details page and remove a resource named "Dependencies", does it make any difference?
2. Have you tried recreating the website and see if you are able to load the Node Details much faster?
3. If you have antivirus, are the SolarWinds folders already excluded from it?
Cancel
Vote Up 0 Vote Down

Cancel
0 dragoon231 over 7 years ago in reply to neomatrix1217

This is what did it more than likely. I did other things that may have helped, but I think the first step of clearing the .NET cache did the trick. The what was inside was almost a gig in size. Clearing it, giving it time to re-create/do it's thing and it is screaming when navigating pages.
thank you for your help neomatrix1217 , janene16 , and Diner !
Cancel
Vote Up +1 Vote Down

Cancel