Router internet slow with IPv6 enabled - works fine from clients

#1

Hi!

I’m on TOS 4.0.1 and have been having this strange problem when I enable IPv6 on my router. Basically it works perfectly fine from all clients - even LXC containers on the router, but is extremely slow on the router itself.

A simple opkg update with IPv6 enabled takes around 50 minutes, as opposed to 5 seconds without it.

I tried time curl -6 https://openstreetmap.cz >/dev/null and it timed out after around 2min, but from my workstation it works in under 0.02s.

A few details of my configuration:

  • ISP is Hyperoptic in the UK
  • Fiber to the Premises
  • Direct ethernet connection via DHCP and DHCPv6
  • /56 address space reserved for my devices

The way I configured IPv6 was by creating a new interface called wan6 bridged to eth2:

config interface 'wan6'
	option proto 'dhcpv6'
	option ifname 'eth2'
	option reqaddress 'try'
	option reqprefix 'auto'

eth2 gets its public IPv6 address just fine:

root@turris:/# ip a show dev eth2
4: eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc cake state UP group default qlen 532
    link/ether d8:58:d7:00:9e:6e brd ff:ff:ff:ff:ff:ff
    inet 100.64.171.115/24 brd 100.64.171.255 scope global eth2
       valid_lft forever preferred_lft forever
    inet6 2a01:4b00:5321:37b8:9006:88ba:c27e:b7d0/128 scope global dynamic noprefixroute
       valid_lft 81964sec preferred_lft 67564sec
    inet6 fe80::da58:d7ff:fe00:9e6e/64 scope link
       valid_lft forever preferred_lft forever

And I get the following PD on the wan6 interface:

IPv6-PD: 2a01:4b00:9d42:9600::/56

I’m not sure where else I can look as I’m relatively new to IPv6. IPv4 works as expected.

#2

Little bit more context, it doesn’t seem to be DNS related as that’s snappy:

root@turris:~# time dig -6 repo.turris.cz

; <<>> DiG 9.11.10 <<>> -6 repo.turris.cz
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 59990
;; flags: qr rd ra ad; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 1

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 4096
;; QUESTION SECTION:
;repo.turris.cz.			IN	A

;; ANSWER SECTION:
repo.turris.cz.		1730	IN	CNAME	proxy.turris.cz.
proxy.turris.cz.	92	IN	A	217.31.192.69

;; Query time: 0 msec
;; SERVER: ::1#53(::1)
;; WHEN: Fri Oct 11 23:15:43 BST 2019
;; MSG SIZE  rcvd: 79

real	0m 0.01s
user	0m 0.00s
sys	0m 0.01s
#3

hi there. Not sure if this helps, but my ipv6 approach is different & working?
I just created a WAN ( ip4) and set the ipv6 conf on auto. ( advanced>obtain ipv6 adress>auto )
Basically the ipv6 is then made by the OS, and all is running smooth?

I repeated your tests, and all is fine here?

best, Dikke

@turrisMOX:~# time dig -6 repo.turris.cz

; <<>> DiG 9.11.10 <<>> -6 repo.turris.cz
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 47362
;; flags: qr rd ra ad; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 1

;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 4096
;; QUESTION SECTION:
;repo.turris.cz. IN A

;; ANSWER SECTION:
repo.turris.cz. 244 IN CNAME proxy.turris.cz.
proxy.turris.cz. 300 IN A 217.31.192.69

;; Query time: 71 msec
;; SERVER: ::1#53(::1)
;; WHEN: Sat Oct 12 08:51:05 CEST 2019
;; MSG SIZE rcvd: 79

real 0m 0.28s
user 0m 0.02s
sys 0m 0.00s

@turrisMOX:~# time curl -6 https://openstreetmap.cz >/dev/null
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 36680 0 36680 0 0 62064 0 --:–:-- --:–:-- --:–:-- 80262
real 0m 0.62s
user 0m 0.09s
sys 0m 0.01s

#4

Thanks for your reply but I have no such option when DHCP is selected as the protocol. It is there for PPPoE but not for DHCP and other protocols.

#5

Hello,

this looks like your ISP is applying some traffic control to the IPv6 address of your router – maybe to protect their infrastructure. Since IPv6 works properly for connected devices, which use prefix 2a01:4b00:9d42:9600::/56, you may try to disable the global address 2a01:4b00:5321:37b8:9006:88ba:c27e:b7d0/128 on the WAN interface by setting:

option reqaddress 'none'

This would make your WAN link unnumbered and force your router to use the LAN adddres for its own communication with the Internet.

#6

Thank you for your suggestion.

Gladly, I didn’t have to do anything and the issue sorted itself out without intervention on my part, so I guess it really was something on the ISP side.

#7

If it happens again, feel free to write us for re-open.