IPv6 BNG issues - things just got real

[#1]

So yesterday I upgraded to an ac68 from the ac66.
It turns out asus on the ARM devices (ac68 is, ac66 is not) forcefully disable file caching, by sending a 1 to drop_caches variable frequently.
When I connect to a BNG gateway the dhcp6 process on asuswrt goes crazy, it stops and starts very frequently multiple times a second clearly malfunctioning, ASUS staff, john (maintainer of fork) and merlin have all looked into this with no clue what plusnet's gateway equipment is doing.  On the ac66 it didnt create a performance impact, it just simply flooded the log with errors.
On the ac68 because there is effectively no caching, i/o goes through the roof on the router and one of the cpu's gets locked into syscalls.  The only way I can prevent this is either been at hand and forcefully killing the dhcp6 master process, or disabling ipv6.  Hopping to a non BNG gateway restores normal behaviour.
So this problem is now bad enough that if I get dumped onto a BNG and either I am not around or I dont notice, it may burn out my router.  Currently we are looking to find a way to either enable caching again on the ac68 (I suspect asus disabled it for a reason tho, I think simply due to memory pressure with the new features) or I make a script that when it detects a out of control dhcp6 process it kills the master process.  MAking a script to auto hop when BNG is detected is also an option.
Really hope plusnet sort this out, because the problem is still dragging on.

[#2]

Would I be right in assuming this only happens when you're using your IPv6 test credentials?

[#3]

I assume so since the ipv4 accounts dont have ipv6.
I am not keen on trying out my ipv4 account given I have just done the effort to get onto a non bng gateway that isnt congested 2 mins ago

[#4]

Reason I ask is due to gauge the severity of the issue. If it's just the IPv6 accounts that exhibit these symptoms then it's something affecting less than ~20 people (where there's a workaround). If it was to affect IPv4 accounts as well then we'd have a much larger problem on our hands.

[#5]

If it affected everyone I expect you would have some more posts about it.
I can tell you if I ignore the problem it affects the router so bad alot of stuff feels laggy as one cpu core is tied up by the dhcp6 process.

[#6]

pressure is off.
I told the dev to give up yesterday as my problem isnt his doing, to my surprise he left me a message earlier tonight he said he solved the caching issue and did a new firmware, the new firmware has reenabled caching so the router no longer sucking up cpu cycles on the bng's.