EAPs show disconnected when using NSSM service
My post was originally building onto @suh4ail 's post found here:
I'm making a new post because that one was getting a little old and most support forums prefer to start a new topic.
We seem to have the same symptoms, and I'm hoping someone has some insight. @Bebe1980 , I'm tagging you here too because after searching for days/weeks in these forums, it's possible you may have had different symptoms but similar reasons for them. At the very least, you seem to have an understanding of how to manipulate the files to make things work the way they need to.
Anyway, here goes:
My EAPs are showing as disconnected when, and only when, I start the java and mongod services using my NSSM-created service
When I start the service using the main program desktop icon, they are connected fine.
In both cases, I can log into the management console and see everything; it's just that the EAPs are showing a disconnected state when the services are started via my NSSM one.
To clarify, here are some knowns:
1) When I use the desktop icon and let the program start the service, my APs are connected and I can manage them.
2) When I run my service, Java and Mongod are both confirmed to be running. I can log in to the management console, but the devices show disconnected.
3) This does not affect connectivity of clients to the access points. Even with the NSSM-made service running, clients can access the network through the access points.
4) When using the desktop icon to start the service and reach the management console, the browser address is: http://localhost:8088. Then it changes to https://localhost:8043 When I use my NSSM-made service and open the broswer, I use 127.0.0.1:8043 The IP address of my server also works. I only bring this up because you did. IN any case, the address I use in the browser does not make a difference for the APs connecting. It's all about which method is used to start the service. In other words, if I let the desktop icon start the service, I can access the management console from any of the addresses and the APs show connected. And vice-versa.
5) I've been running my controller as a service for probably five years now, and editing the service parameters as needed as the controller has upgraded.
6) I've made notations in different logs to check differences in startup methods. in the tp-link/controller/logs folder is logfile "mongod.log" When I compare the log entries from NSSM-made service startup method versus the desktop icon startup method, they are very close to the same, but there are differences. I haven't parsed everything yet, but the NSSM-made service definitely seems to have more entries and has some errors in it. I think this will be my next avenue of pursuit.
7) I am running Java verision 1.8.0_301 The guide on the forums here discusses a previous version, but it's still 1.8.0 and I don't think it's the main cause. I still might remove java and reinstall the older version for testing.
8) The service guide on the forums starts out great. But when it gets to the part about the alternate method, nothing is explained well and it doesn't match what I'm seeing.
9) I'm using EAP660HD APs. I have two of them. I've taken my older EAPs out of service for now so they are not part of the equation.
a) The server.log file I think will be key. Something is going on in the service startup that is not- allowing AP connections properly and there may be clues in the log.
b) I feel like there's something missing in the java calls when we create the service, like maybe we need to add arguments or environment tables. But I'm not fluent enough in any of this to know with confidence what I'm talking about. I might get lucky and stumble upon something, but it's not likely.
c) I might start reading into the .jar files.
Windows Server 2012R2.
Omada Controller 4.4.6, downgraded to 4.4.4 with same symptom
EAP 660 HD x2