When I try and change the polling engine of an agent node by editing the node settings and pointing it at a new polling engine it doesn't seem to work. Orion seems to acknowledge that the node is on the new polling engine; however, the node then shows as down. I had to go edit the agent settings on the node itself and in there it still showed it pointed at the old polling engine, I had to update this to reflect the new polling engine also before the change was working properly.
It seems like I should be able to do this without having to log into each node, is this not the case?
Thanks in advance for any suggestions!
Is it possible that one or both APE's is not allowed to connect to the main poller over TCP Port 17777? Similarly, the main Orion server must also be able to connect to all APEs directly over TCP Port 17777.
aLTeReGo I wanted to let you know that I opened a support case on this and was informed that for agents using the Agent Initiated communication model that you can't change the polling engine via the web console, they need to be manually changed on the node. I was not satisfied with this answer so the support guy (who was great to work with) confirmed this answer with two AE's.
Can you confirm if this is accurate?
If that is accurate do you guys have documentation on the proper steps to take to change the polling engine for an agent running in Agent Imitated mode?
Is this possible via the SDK? Its a pain to do this manually for each node. The only way I can think to do this en masse is to change the polling method, update the table and change the polling method back. The advice in Polling Engine Load Balancing · solarwinds/OrionSDK Wiki · GitHub doesn't push the change to the agents.
Hey @aLTeReGo, these links go to a general page and search is not yielding favorable results.
can you put the actual titles of the linked videos or update the links...I'm banging my head against this currently.
Yeah, I get that, but the challenge is that we have ~3000 total devices, of which about 50 are managed by an agent.
I'd like to be able to do a
where nodeid %2 = 0;
where nodeid %2=1;
Or something similar using the SDK. If I run that now, my devices that are managed w/ an agent whose poller changes show as down. I suppose I could craft my where clause to leave those nodes out and do them manually, but its a bit of a pain. Was hoping for a more automated way.
Funny thing you say that, less than 5 minutes ago I got a response on my support ticket I have opened on this and he pointed out the same thing to me when I noted this post. I guess this means I will be waiting for version next before I try and move several dozens of nodes all at the same time.
The main Orion poller can definitely connect to all of the APE's on TCP Port 17777. The APE I was moving the node from can communicate to the main poller on TCP Port 17777. The APE that I was moving the node to can also communicate to the main poller on TCP Port 17777; however, it does so over a NAT IP. If this were causing problems shouldn't I have expected to see other more significant issues?
I apologize for the delayed response byrona. How is the Agent configured? Is it in Agent Initiated (Active) mode, or Server Initiated (Passive) mode? If the Agent is running in 'Active' mode, can you confirm that the machine where the agent is installed can properly resolve the hostname (short name, not FQDN) of the polling engine you're trying to move the agent to? If it cannot, that would explain why the Agent keeps rehoming back to the last poller it was able to successfully communicate with. This is a safety measure built into the Agent to prevent them from becoming orphaned should they not be able to communicate with the desired polling engine.
aLTeReGo I added a host file entry for the short name of the polling engine and was able to ping it using the short name to confirm it was resolving properly. When I attempted to move it to a new polling engine it failed again. The problem seems to be that it's updating in the Orion database that it's on the new polling engine but the node itself isn't getting updated. In the screenshot below the Orion Agent settings continue to show the old polling engine but in the Orion WebUI it shows the new one.
byrona, have you confirmed that each of the Polling Engines can resolve eachothers hostnames (shortname, not FQDN)? E.G. The main poller must be able to resolve the hostname of all additional polling engines. Similarly, the Polling Engines must be able to resolve the hostname of the main poller.
aLTeReGo I am running the agents in Agent Initiated (Active) mode. My expectation is that they likely can NOT resolve the hostname of the poller. If what you are saying is correct I should be able to test this by putting in a host file entry allowing it to resolve. I will test this and let you know what I find.
Also, are you able to comment on the accuracy of the information I was provided by Support indicated below?
SolarWinds solutions are rooted in our deep connection to our user base in the THWACK® online community. More than 150,000 members are here to solve problems, share technology and best practices, and directly contribute to our product development process.