Hub 2 requires factory reset roughly every 2 weeks

[#1]

I notice from the event log, that it looks like you have IPv6 enabled ?

IPv6 is not publicly available from Plusnet yet. Whilst having it enabled SHOULDNT be a problem, I'd be inclined to try disabling it...

[#2]

I have disabled ipv6 in the past but after factory resetting due to drops it gets turned back on by default.

[#3]

What does your web server actually provide to your family ?.

For example -

• Does it stream video to your family members when away from home ?
• Do you host large files that are shared between family members ?
• Do your family use your web server as cloud backup for their mobile devices ?

[#4]

What does your web server actually provide to your family ?.

For example -

• Does it stream video to your family members when away from home ? - No
• Do you host large files that are shared between family members ? - Yes
• Do your family use your web server as cloud backup for their mobile devices ? - Yes

I have a navidrome music server, a filebrowser service providing webdav file access, i run syncthing with multiple mobile phones, i run immich photo server, joplin notes, and stream a cctv/webcam.

All is behind a nginx reverse proxy with ssl from lets encrypt, and backed by fail2ban.

[#5]

What does your web server actually provide to your family ?.

For example -

• Does it stream video to your family members when away from home ? - No
• Do you host large files that are shared between family members ? - Yes
• Do your family use your web server as cloud backup for their mobile devices ? - Yes

I have a navidrome music server, a filebrowser service providing webdav file access, i run syncthing with multiple mobile phones, i run immich photo server, joplin notes, and stream a cctv/webcam.

All is behind a nginx reverse proxy with ssl from lets encrypt, and backed by fail2ban.

[#6]

---

@fozid wrote:

The other funny thing is that during these drops, I am not able to access the hub2's webui, even from my pc which is also connected through ethernet.

---

Isn't that a known symptom of the Hub2 when the upload path gets totally overwhelmed with traffic, then the Hub2 GUI and connectivity becomes unresponsive due to bufferbloat.

If the LAN side web server was sending a large file (such as video) out to the internet, then transmission overload of the Hub2 could easily occur.

@fozid - Can you use your Ngnix reverse proxy to set a rate limit on the outbound bandwidth ? - if so perhaps set it to 60% of your internet connection's normal upload speed.  Having done a quick Google, there are a lot of YouTube instructional videos on how to rate limit with Ngnix.

.

[#7]

Looking at the accepting inbound port forward rule connections I'd be asking myself who needs top be connecting to your web server at during the night hours. The inbound connections are constant, so many the Firewall event log is truncating

02:07:28, 30 Jul. FWL Port Forward Server(192.168.1.4) TCP 443 accepted a new connection from 43.157.170.13
<<<<<<<<<<<<<<<<<<<< Limit of FW log >>>>>>>>>>>>>>>>>>>>

Take a look at the event log at these messages, I'm interested why you're getting a connection from 45.135.232.92 (there are many)

08:28:00, 30 Jul. FWL Port Forward Server(192.168.1.4) TCP 22 accepted a new connection from 45.135.232.92

https://en.ipshu.com/ipv4/45.135.232.92 

https://en.ipshu.com/ipv4/43.157.170.13 

Do you really need port 22 open, and why?

[#8]

cool, i have added a bandwidth limit of 25mb/s to my server. Max is ~30mb/s.

[#9]

port 22 isnt directly open, there is a port forward on a different port externally set to 22 internally. and yes, i need 22 access remotely. I didnt have fail2ban setup correctly. this should resolve the multiple attempts from random IP's, although they are all trying to access a non existent user (root)

[#10]

---

@fozid wrote:

... great tip on the Broadband Quality Monitor. I have set that up 👍

---

@fozid 

Can you share your BQM graph so we can check if anything unusual is going on there ?.

Go to https://www.thinkbroadband.com/my-profile/broadband-quality-monitor 

Click on [Share Live Graph]

Copy the URL in the "Direct Link"

then paste that URL in to your reply here.

.

[#11]

https://www.thinkbroadband.com/broadband/monitoring/quality/share/90bb06326812334602379177d3493c1b2f138644

[#12]

Above link reformatted to make it more easily visible -

[#13]

---

@outcast wrote

Isn't that a known symptom of the Hub2 when the upload path gets totally overwhelmed with traffic, then the Hub2 GUI and connectivity becomes unresponsive due to bufferbloat.

If the LAN side web server was sending a large file (such as video) out to the internet, then transmission overload of the Hub2 could easily occur.

---

That might well explain the symptoms reported on the opening post …

“When we experience issues, the blue light remains solid on the hub2, implying it is connected to the internet, but the devices connected on WiFi show no internet.”

It is not clear  what “show no internet “ means, but windows devices run some form of probe to affirm internet connectivity.  If a massive US transfer is running, other devices might not get a look in for their (low priority) probes, giving rise the the appearance of there being no internet connectivity.

i do not though understand how a factory reset achieves more than a power cycle.  There is some possibilities of memory leakage which would be fixed by a power cycle.

My Hub Two runs for months on an unbroken xDSL session without apparent issues.  A factory reset would require a rebuild of firewall and other settings ..  is there something therein which results in ‘better’ operation … or is this just positive coincidental reassurance?

[#14]

---

@Townman wrote:

---

i do not though understand how a factory reset achieves more than a power cycle.

---

My thoughts about that was that it is just a matter of different timing.

If for example you were a family member remotely trying to (say) stream a video from the web server, if the stream stopped or was buffering badly,  then the chances are that a router power cycle would complete sufficiently quickly, that the download might resume, perhaps at lower resolution due to degraded connection or whatever, and the fail cycle might repeat a few times with varying degrees of severity, or the remote user moves on to try something less demanding of incoming traffic.

However, if the router has been factory reset and adding the port forwards etc, then the time to establish the connection would be so long, the the remote user would have given up trying, and therefore the router's upstream would be free of upload traffic, and therefore appear to be working.

[#15]

when i get the issue again, i will generate a fresh event log from the router, and upload it here. 

i have added rate limiting and a fairly strict fail2ban config, so that should limit the potential issues.

until then its hard to look at what’s causing the issues.

thanks for everyone’s input so far 👍

[#16]

---

@fozid wrote:

when i get the issue again, i will generate a fresh event log from the router, and upload it here. 

Given that the BQM  (reposted above) is the last 24 hour scrolling plot of your connection's latency

if you do have the need to post new event logs etc, can you also include the static BQM snapshot of the day that the associated event log covers, so the new chart doesn't disappear from sight and be compared at a later date with any events that occurred.

.