Intermittent hub freezes cured by access from wired device

[DamnMachine2]

I am experiencing random intermittent hub freezes, which seem to be cured instantly when a wired device connected directly to the hub accesses the internet.  This has been going on for years, and I've been swapping components one by one without finding the cause.  I'm currently baffled, so asking for help from the hive mind.  

It happens on average once a month.  But sometimes we get two in two days, sometimes none for two months.  

Current Configuration: 

Volt M200 broadband, GU post code. 
Hub 5 - installed only 30 days ago
Port 1 on hub connects to a four port Ethernet switch with a laptop, fileserver, and printer
Port 2 on hub connects to a DecoX55 mesh wifi hub, with wired backhaul to two other DecoX55s, each of which have a two port switch built in from which I connect to a wired device.
We use the Deco mesh wifi, so wifi on the Superhub is switched off to avoid any possibility of frequency conflicts. 

Symptoms: when the error occurs, wireless devices can still connect to the Deco, but the Deco maintenance utility reports it has lost internet access.  A laptop connected via the Ethernet switch also says it has no internet connectivity.  

Cure: Turning the Hub5 off for 5 mins then on again will bring back service.   But quicker than that, if I connect a laptop directly to one of the hub ports and open a web site, everything else springs back to life instantly.  (The location of the hub prevents me from leaving any devices permanently directly connected to it.) 

While directly connected, I have had a look in the hub logs and found entries like these around about the time of the failures:

Warning	RNG-RSP CCAP Commanded Power in Excess of 6 dB Below the Value Corresponding to the Top of the DRW;CM-MAC=xx:xx:xx:xx:xx:xx;CMTS MAC=xx:x:x:xx:xx:xx;CM-QOS=1.1;CM-VER=3.1;
Critical	Started Unicast Maintenance Ranging - No Response received - T3 timeout;CM-MAC= xx:x:x:xx:xx:xx;CMTS-MAC= xx:x:x:xx:xx:xx;CMQOS=1.1;CM-VER=3.1;
Error	DHCP RENEW WARNING - Field invalid in response v4 option;CMMAC= xx:x:x:xx:xx:xx;CMTS-MAC= xx:x:x:xx:xx:xx;CM-QOS=1.1;CMVER=3.1;

(MAC addresses masked out). 

I'm curious to know if these could be relevant to the failures I am seeing. 

This pattern of random intermittent outages has gone on for many years.  Before the Deco mesh Wifi I used a second wifi hub with better coverage than the Virgin hubs - with similar behaviour.   I've also tried changing out the Ethernet switches.  At once stage I was using powerline connections to distant wired devices and thought they were the problem.  But they are all gone now.  My most recent theory was that my last hub, a Virgin Superhub3, must be faulty or old.  But a brand new Hub5 is displaying the same behaviour. 

All suggestions very much appreciated. 

[Adduxi]

Are the Deco units purely setup as Access Points only?  Is the Hub 5 the only DHCP server on the network?

[DamnMachine2]

Thanks for looking at this. 

Yes.  Deco Operation Mode = Access Point.

And Deco Smart DHCP = Disabled.

[DamnMachine2]

And 
ipconfig /all | find /i "DHCP Server" gives
DHCP Server . . . . . . . . . . . : 192.168.0.1

which is the Hub 5.

[Adduxi · Accepted solution]

As far as I understand from the Internet, RNG-RSP occurs when the Hub has a disconnect event. It could be a faulty connection somewhere in your circuit. I would setup a BQM and post your Hub's stats for comment, i.e. power levels, etc.  Read this blog, it's a very interesting timeline around resolving a RNG-RSP fault. Solving Intermittent Cable Modem Issues (duckware.com)

It may be nothing to do with your network setup, but more an ongoing long time fault in the VM cables? 

[DamnMachine2]

Thanks.  That's good advice.  I was going to ask you what to look for on the BQM monitor, as I've had them running and not spotted any particular pattern when the outages occur. 

But then I just went and looked at my latest one.  It looks extraordinary to me: 

Surely there shouldn't be all those dropped packets?  The network itself seems to be performing fine.  No outages during the time period shown. And I get 100 Mbps download and 25 Mbps upload in speed tests, just as I would expect.

[Client62]

Cracking BQM graph !

As ever it tell us zip about where or why the pings were discarded. Or could it be not even sent !

[jbrennand]

Out of interest... why on earth are you not using the VM Hub in modem mode with the Deco doing all the heavy lifting?  The X55 modem and wifi are a quantum jump better than that in any VM Hub.

Take control of your system and allow the Hub to work as a modem only.

[DamnMachine2]

I did use modem mode, about six configurations back, maybe 10-15 years back.  I stopped using it because when I encountered problems with the outbound Virgin connection, it was very difficult to diagnose them.  That was probably two hubs ago, though.  I could be persuaded back again if the Hub5 still gives you logs and connection stats even when in modem mode? 

[DamnMachine2]

I just looked back over the last month's BQM monitors.  Those dropped packets started appearing on the day the Hub5 was installed.  The red is there most of the time most days since then.