Weird Network Issue

I have a very weird issue...  I have my MyServer and Home Assistant running on my QNAP NAS.  Everything works great.  I have my Global Cache devices on a separate network segment with a route defined in my Firewalla Router.  Everything works fine and my MyServer can communicate to the separate network easily.  

Here is the weirdness...  I can ping all devices via my PC and access their config screens.  If I try to ping my GC-100 via the Myserver or HA, the ping hangs and blocks communication to the device from Myserver or HA.  This behavior happens only with the GC-100 and follows it (meaning, if I change the IP, the same thing happens with the new IP).  This behavior does not happen with the other Global Cache devices on that network.

I discovered this because the GC-100 will disconnect and reconnect its ports at random times during the day.  I tried to setup the Ping Integration on HA to keep an eye on the device.  The Ping Integration shows the device as disconnected (but I can ping it from the PC and access the config screen).

I have a Ubiquti ER-3 as the router between the main network (192.168.2.x) and the (192.168.6.x) network.  It doesn't report any errors.

The reason for the separate network is that the Global Cache devices and the Vrroom would drop off the main network.   Apparently on a larger network, the interfaces don't like all the noise and shutdown, both Global Cache and HdFury suggested putting them on a separate segment.

***Short Version***

Pinging the GC-100 works from a PC, but from a virtual machine on a NAS fails.  Normal communications work fine.

Any ideas?

 

Weird. By any chance do you have QuFirewall enabled?

No...not installed. One more weird thing, I can ping it via Windows devices but my Linux workstations fail. If I run Ubuntu on the Windows 11 laptop, the pings fail from it....

Makes no sense. I can ping all the other devices on that segment and can hit the GC-100 with the web browser to configure it.

????????????

Getting to the root of the issue... here is what is happening in the log.

30/12/2025 08:02:44.02 · DEBUG · GenericSerial.NetDisconnect - Device (91) has disconnected.
30/12/2025 08:02:44.02 · DEBUG · GenericIR.NetDisconnect - Device (89) has disconnected.
30/12/2025 08:02:54.02 · DEBUG · GenericSerial.NetConnect - Device (91) has connected.
30/12/2025 08:03:44.03 · DEBUG · GenericIR.NetConnect - Device (89) has connected.

89 = The GC-100 ID in the GenericIR Driver
91 = The GC-100 Second Serial Port (in both the IR and the Serial driver, deactivated in the IR Driver)

I have been working with Global Cache Tech support and they are telling me that the GC-100 supports a single TCP 4999 connection. If a second connection is tried, it will drop a connection and let it connect.

This "feels" like the issue because the GC-100 "lives" in two drivers (GenericIR and GenericSerial).

This behavior is consistent between GC-100 devices (I have two of them).

Global Cache Tech support is working on this and I will update the thread with any updates.

This isn't a huge deal because the port drop-off is sporadic and only lasts for about a minute. I only noticed it because when I was shutting down my HT, one of the front powered tower speakers didn't turn off properly.

Here is the formal response from Global Cache Tech Support...

"The log entries strongly suggest a connection arbitration issue caused by both the GenericIR and GenericSerial drivers interacting with the same GC-100. While the device exposes IR, serial, and relay functions, it effectively supports only a single active control session. When the two drivers independently manage their connections, the GC-100 may determine one session is stale and reset it, which results in both drivers disconnecting at the same time and then reconnecting in a staggered sequence. This aligns closely with the timestamps in your logs and explains why the behavior occurs even when the system is idle."

I wonder if there is a magic config that would mitigate this issue (for example, deleting the references to the non IR ports in the GenericIR driver vs making them inactive.)  I'm going to bring my backup device online to muck around a bit. 

I mucked around a bit with the GC-100 and found some odd behavior. You can ping it via Windows but the ping doesn't respond via Linux... No big deal. The disconnects and connects also happen when the GenericSerial driver is configured and the GC is not in the GenericIR driver. This shows that it is not the two drivers fighting with each other.

What I can determine is that the GenericIR driver will cause the base device to cycle throughout the day, the GenericSerial driver only impacts the serial ports.

The problem is that the disconnect and reconnect can take up to a minute and when it is happening, any commands sent to that port fail.

I switched out one of the Serial ports to a IP2SL device and it has been rock solid. I might just order another IP2SL and call it a day.

FYI - I have two GC-100's and the behavior is the same in both devices (showing it's not a faulty device). Global Cache also said the firmware was the latest.