Atom self healing

[M2macro]

Wifi cutting out momentarily. Returning this notice ‘NOTICE ATOM is restarted due to Kernel/oops panic as part of Self Healing Mechanism’ 

CPU issue? Can this be fixed? Customer service team no use. Called 3 times now and each time have suggested something different that did NOT help.

[jaysabherwal]

Makes sense, phone customer service have been an absolute pain, Ive told them the exact issue so many times and they just ramble on about some other thing.

[Sephiroth]

---

@jaysabherwal wrote:
Makes sense, phone customer service have been an absolute pain, Ive told them the exact issue so many times and they just ramble on about some other thing.

---

... and that's what the Forum Team here will do.  Anything but acknowledge the fatal flaw of the Hub 3.  If you're lucky, they'll offer to send a tech out to you who might swap the hub - prolly for another Hub 3.

If you're unlucky, they'll say they've looked at your circuit data and can see nothing wrong, which would be utter tripe.

Keep us posted.

[IPFreely]

FYI It's actually the Atom SoC itself that's being restarted. The kernel crashes and to recover from that crash the CPU it's running on has to be restarted. To kill the kernel usually requires either hardware to be at fault or a device driver that's running at kernel level to behave improperly.

https://en.wikipedia.org/wiki/Kernel_panic

The Hub can stay connected to the cable network as that's separate hardware connected to the SoC over some sort of bus but anything dependent on the Atom being present all the time to run sees interruption.

[Adduxi]

Yes, correct on all counts.  The Puma flaw was “fixed” I believe by moving some of the CPU load to the Wi-Fi Chipset.  I’m guessing the ATOM CPU still gets overloaded at times. 

[IPFreely]

They offloaded some traffic from the dedicated packet forwarding hardware to the SoC rather than the other way around.

I'm not clear on the exact details but this is why some UDP traffic is heavily limited on performance. It's going though an Atom SoC rather than a network ASIC. 

I have seen the same issue on Surfboard modems but thanks to it being a modem it's not really noticed: no WiFi requiring the SoC to break, nearly everything forwarding as usual as it doesn't involve the SoC. I assume the same holds for bridge mode on VM Hubs. 

[jaysabherwal]

Got on the phone with them again yesterday, this time went through retentions to attempt to get a hub 4 but they said that unless I upgrade to 1Gig (by doubling my bill cost ofcourse) I would not be able to.

They transferred me to technical help and the rep was helpful and organised a technician for this morning (unexpectedly quick), got a new Hub 3, will monitor it for the next few days.

[Rodinbangor]

Don't expect any change.

I've had 3 x Hub3, hasn't made the slighest difference. 

Its simply a piece of junk that isn't suitable for the job.

The sad thing is that Ofcom don't care, & let VM get away with it instead of suspending their licence.