5G hub (Greenpacket Y5-210MU) dropping wired connection after a few days every few minutes

[JohnR]

First, it seems I'm not alone: https://www.ispreview.co.uk/talk/threads/drop-out-problems-with-new-three-outdoor-hub-y5-210mu.42890...

tl;dr - I've only had the Three 5G broadband package for 1 week. I didn't alter the 5G outdoor hub  configuration, which is running firmware 130.00100.113.024. After 6 days of flawless operations with minimum 300/20 Mbps downlink/uplink, the 5G outdoor hub started closing the ethernet connection intermittently every few minutes, and the only fix (so far) is a reboot of the hub. I'm willing to help the Three engineers debug this (I have networking experience), but if it can't be properly fixed (firmware update?) or if they aren't interested, then I'll have to cancel the contract as this is a deal breaker. 

Details:

The outdoor hub (Greenpacket Y5-210MU) seems to intermittently close the ethernet connection. I've debugged this several ways: with a laptop directly connected to the hub via the PoE, via the eero, via its own web GUI (192.168.0.1).

I've only had it for 1 week. Worked fine for 6 days. Then last night I noticed intermmittent dropouts. Pinging google.com continuously revealed the dropouts last 30-60 seconds, and happen every few minutes, from 5 to 10, but especially if there is any network activity from some connected client (phone, laptop, etc).

I used several methods to discover this:

1. the web gui's advanced > system > ping (which pings directly from the hub, independently of any connected router/eero/device); this always works fine, so I know the 5G connection is working 
2. a laptop connected directly to the hub; when it happens the laptop OS reports ethernet disconnected
3. the supplied eero 6 router; when it happens, the eero's led turns red, and the eero app says "Internet Offline, Wired Connection Not connected". 
4. The syslog logs don't show anything suspect, which to me suggests a memory leak of a process running on the hub (syslog logs downloaded from web GUI > advanced > system > maintenance > syslog > download) 

I waited for almost 24h, during which it lept happening continuously, then I rebooted the Greenpacket outdoor hub and voila ... came back to normal.

From experience with modems and routers, this is a symptom of a memory leak of a process that runs on the Greenpacket hub, which builds up over time until it fills the entire memory, and then all services get affected, causing interruptions everywhere.

Are there others experiencing this?

Are there any Three employees here willing to help with this?

[Brianboitano]

It is now dropping every single day.  At 4:58 AM according to Eero logs that is when the ethernet connection to the hub breaks.  Is anybody from Three in this thread?

[MymsMan]

Have you checked checked the scheduled reboot interval on the hubs advanced settings page?

The default is normally 7days but it sounds as if yours has somehow changed to daily 

Interval can be changed to up to 30 days or disabled 

[Brianboitano]

Yes I have checked it and it is disabled.  It is not rebooting because it never comes back on again unless it is hard rebooted by removing the plug.  It is crashing not rebooting.

I appreciate your attempts to help.

Are Three just ignoring this issue and hoping it goes away?

[MymsMan]

---

@Brianboitano wrote:

Yes I have checked it and it is disabled.  It is not rebooting because it never comes back on again unless it is hard rebooted by removing the plug.  It is crashing not rebooting.

I appreciate your attempts to help.

Are Three just ignoring this issue and hoping it goes away?

---

Very strange that it is happening so consistently at 4:58 - something must be scheduling it!

Does the hub still respond to attempts to access it via https://192.168.0.1/ and does a soft reboot via the reboot button https://192.168.0.1/#/advanced restore connectivity?

If it does then you could use a script like @gavinmcnair's autorestarter  to automatically reboot 

I don't know if Three consider rebooting to be an (in)adequate solution or just haven't assigned the problem to someone with sufficient thechnical prowess to investigate the problem thoroughly.

[sargentdel1]

This was Jonathan’s reply 3 hours ago further up in this thread…

I've chased this up and the team haven't seen similar issues on our test devices, so it seems like it's not a universal issue. I'll chase this up further though.

How often would you say the issues are recurring?

Thanks,
Jonathan

Seems a little hard to believe given the amount of issues that have been reported in this thread alone. 

[ricdacruz]

ask if thy would be willing to swap one of their test devices for yours?