AMP exhibits inexplicable behavior when connecting over inet but no issues over LAN

  • 43 Views
  • Last Post 04 May 2020
Sabinno posted this 02 May 2020

EDIT: After looking at the logs, I realized that my IP address kept changing when connecting, creating session reuse errors.

This was a result of CloudFlare protection, which I am not keen on disabling.

I found the solution in this thread, also on this forum.

If you use CloudFlare, one must enable Login.AllowInsecureMatching=True in AMPConfig.conf.


OS Name/Version: CentOS 8.1.1911 64-bit (on ESXi 6.7U3 host)

Product Name/Version: AMP 1.9.9.2 (running as standalone, not as controller/target)

  • Only one instance has been created. It is a single Minecraft instance with very few changes to the default configuration other than enabling FTP at the ADS level.

Problem Description: When logging in remotely rather than over the LAN, AMP behaves in inexplicable and strange ways. When logging in over LAN, all features of AMP work as intended with zero bugs or issues whatsoever.

  • AMP will not show instances, only showing: "No remote targets have yet been set up. This ADS installation has been setup as a controller, which means it cannot create instances by itself. You must create a Target ADS instance and register it with this controller before creating instances." This is definitively untrue, as it was configured as standalone for the first time just two days ago.
  • Navigating to "User management" shows no users whatsoever, including the logged in user.
  • On Chrome/Chromium-based browsers, AMP forcibly logs out the user after a few seconds.
  • Logging in via various proxy servers produce even weirder issues, such as random UI elements being displayed in place of error messages or instance names.

Steps to reproduce:

  • Successfully connect to AMP on LAN with no issues
  • Disconnect from LAN and use a mobile hotspot connected via ethernet to test remote connectivity
  • Attempt to log in to AMP via contoso.com:8080

Actions taken to resolve so far:

  • Use different browsers (Edge, IE11, Firefox, Opera) - same issue, though Chrome appears to be the only browser that kicks me out repeatedly after login
  • Flush DNS resolver cache (Windows) - no results
  • Empty browser caches on all browsers - no results
  • Delete ALL browser settings on all browsers (or use incognito) - no results
  • Verified correct user permissions from LAN.
  • Used two entirely different devices to connect (a Chromebook and Android phone) on the same hotspot -- no results
  • Used various OS-level proxy servers to connect from all devices - entirely different set of issues occurs (random UI elements placed at top of page, no instances nor error messages displayed, very few options presented)
  • Update OS and all packages - no results
  • Update ampinstmgr and all instances using steps for CentOS from Install/Update page on cubecoders.com
  • Obviously, rebooted several times. Both ESXi host, VM, and the single instance.

Actions NOT taken to resolve so far:

  • Use a different network - I don't have access to another network with which to test yet, as all cafes in my area are closed, I'm working from home, and visiting people is fairly irresponsible at the moment.

Mike posted this 04 May 2020

You may need to turn off the "Require Session IP Stickiness" setting.

Close