Having problem with China websites

[bert64]

That is why I am puzzled.

If we attribute the issue to high ping due to bad routing, then all of us should be affected. But apparently only some SingTel users are affected.

It could be because they might be so unfortunate to go through even worse routing (say through China Telecom) due to different IP address segment or different Taobao server locations.

Simple ping and traceroute may be able to check whether the hypothesis is correct.

Singtel are making 2 separate BGP announcements under 121.7.x, instead of announcing the whole range as a /16 they have split it into a pair of /17 announcements:

https://bgp.he.net/AS9506#_prefixes

Wether both of these prefixes are announced to their various peers equally, or accepted by those peers i couldn't say.
There could also be other problems, such as rogue announcements or localised misconfigurations. The ipv4 address space is small and very messy, it's not uncommon for companies to have used previously unallocated address space internally. For instance there's a lot of companies that use 1.1.1.x as the captive portal page on their wifi etc.
This only manifests itself when someone is actually allocated the address space and tries to communicate with you, for instance someone using a wifi setup like this would be unable to access the cloudflare dns service now using the 1.1.1.1 address.

 

[bert64]

For Taobao, the website front page can be partially loaded, some resources are failed to load. If you do search in Taobao, will take a while to load or cannot load at all.

If you use the developer tools option in chrome or firefox you will see the individual requests it's making for each resource on the page. If those resources are loaded from servers which are slow/unreachable this kind of problem will occur... It seems the main static content of the page is cached locally so should always load quickly.

Based on the results earlier i've tried from servers in a number of different locations (but none in china) and generally always have very poor performance with high latency and packet loss to "s.taobao.com" as used for the search option...

So i would say it's most likely a problem with their own international transit.

 

[SeanPC]

You can try but that may or may not work.

If you have an older SingTel router, just switch to that old router and then put back the current router. This will sure work in terms of changing the dynamic IP address. Make sure you do not switch more than this, if not you will get into trouble.

Or if your router supports changing the MAC address, that will work as well.

Still in this case, it may or may not related to your IP. To confirm if it is SingTel Fibre problem or not, you can also try using your mobile phone data connection to see if that works or not.

BTW, my wife is a heavy Taobao user at home and she has no issues with SingTel Fibre. But she is using the mobile app on iPhone. You may want to check if you can use the mobile app over wifi.

Ref: same issue
https://forums.hardwarezone.com.sg/...roblem-accessing-china-website-6059148-3.html

Solution: you can try changing DNS server or VPN or things like Cloudflare 1.1.1.1 mobile app which changes your DNS server.

can use the mobile app on wifi but not the website. same like pc. the website cannot access. mobile data both ok. i also having problem with bitorrent.

 

[xiaofan]

can use the mobile app on wifi but not the website. same like pc. the website cannot access. mobile data both ok. i also having problem with bitorrent.

I see. So this indeed points more to the problem with "s.taobao.com" with SingTel Fibre.

I will suggest try China VPN to see if that helps. I have paid AJS 爱加速 VPN myself which has many servers across China.

You can even try non China VPN as well, say those with HK server.

At least you have a workaround to use mobile app over WiFi.

Bit Torrent is a different issue. SingTel will throttle BT traffic.

 

[miloaisdino]

Singtel are making 2 separate BGP announcements under 121.7.x, instead of announcing the whole range as a /16 they have split it into a pair of /17 announcements:

https://bgp.he.net/AS9506#_prefixes

Wether both of these prefixes are announced to their various peers equally, or accepted by those peers i couldn't say.
There could also be other problems, such as rogue announcements or localised misconfigurations. The ipv4 address space is small and very messy, it's not uncommon for companies to have used previously unallocated address space internally. For instance there's a lot of companies that use 1.1.1.x as the captive portal page on their wifi etc.
This only manifests itself when someone is actually allocated the address space and tries to communicate with you, for instance someone using a wifi setup like this would be unable to access the cloudflare dns service now using the 1.1.1.1 address.

but there does not seem to be a technical issue with singtels routing to taobao servers.. (although the latency is high with packet loss). the routing looks quite expected to me, its already optimised by going through the hk route. ive seen packets being routed to china via the US (zayo) before that result in even higher packet loss on singtel! given that along the route the latency to the closest offload node (china unicom) is only 110ms, the high latency is due to singtel not paying for the higher quality transit into china (CN2) which is a commercial decision. traffic into china on the "regular" route has to pass through DPI in the GFW which is always congested

 

[xiaofan]

From the other similar thread.

Interesting that SingTel used this brutal force method to sort out the issue. They should try assigning a new IP address to see if the issue goes away.

Hello Xiaofan,

After I replaced ONR, issue solved.

I believe the previous IP was blocked, new ONR has new IP, so all applications and website go through.

For your information.

 

[miloaisdino]

From the other similar thread.

Interesting that SingTel used this brutal force method to sort out the issue. They should try assigning a new IP address to see if the issue goes away.

haha. singtel cant reassign ip address since the onr mac addresses are locked.. the only way for a new mac is to replace the onr. singtel also cant clear dhcp leases!!

 

[xiaofan]

haha. singtel cant reassign ip address since the onr mac addresses are locked.. the only way for a new mac is to replace the onr. singtel also cant clear dhcp leases!!

Okay the ONR is limited.

But why SingTel can not clear DHCP leases? What kind of backward DHCP server are they using?

 

[bert64]

From the other similar thread.

Interesting that SingTel used this brutal force method to sort out the issue. They should try assigning a new IP address to see if the issue goes away.

Problem is that IP is going to end up later reassigned to another customer, who will then have the same problem... Such is the polluted mess of ipv4.

 

[xiaofan]

from my azure server!

azureadmin@azure:~$ ping apicn.hualaikeji.com
PING apicn.hualaikeji.com (120.92.43.182) 56(84) bytes of data.
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=1 ttl=44 time=74.0 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=2 ttl=44 time=74.3 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=3 ttl=44 time=74.3 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=4 ttl=44 time=74.3 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=5 ttl=44 time=74.3 ms
^C
--- apicn.hualaikeji.com ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4006ms
rtt min/avg/max/mdev = 74.068/74.288/74.393/0.319 ms

Indeed azure is pretty good, testing from my one year free tier azure server located in Singapore.

xiaofan@azuretest:~$ ping -c 6 s.taobao.com
PING v6-sh-sz-zb.wagbridge.alibaba.taobao.com.gds.alibabadns.com (106.11.93.6) 56(84) bytes of data.
64 bytes from 106.11.93.6 (106.11.93.6): icmp_seq=1 ttl=19 time=70.6 ms
64 bytes from 106.11.93.6 (106.11.93.6): icmp_seq=2 ttl=19 time=70.7 ms
64 bytes from 106.11.93.6 (106.11.93.6): icmp_seq=3 ttl=19 time=70.6 ms
64 bytes from 106.11.93.6 (106.11.93.6): icmp_seq=4 ttl=19 time=70.8 ms
64 bytes from 106.11.93.6 (106.11.93.6): icmp_seq=5 ttl=19 time=70.6 ms
64 bytes from 106.11.93.6 (106.11.93.6): icmp_seq=6 ttl=19 time=70.6 ms

--- v6-sh-sz-zb.wagbridge.alibaba.taobao.com.gds.alibabadns.com ping statistics ---
6 packets transmitted, 6 received, 0% packet loss, time 5007ms
rtt min/avg/max/mdev = 70.644/70.700/70.842/0.274 ms

xiaofan@azuretest:~$ ping -c 6 apicn.hualaikeji.com
PING apicn.hualaikeji.com (120.92.43.182) 56(84) bytes of data.
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=1 ttl=44 time=72.1 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=2 ttl=44 time=72.3 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=3 ttl=44 time=72.2 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=4 ttl=44 time=72.2 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=5 ttl=44 time=72.3 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=6 ttl=44 time=72.4 ms

--- apicn.hualaikeji.com ping statistics ---
6 packets transmitted, 6 received, 0% packet loss, time 5006ms
rtt min/avg/max/mdev = 72.175/72.297/72.428/0.090 ms

 

[xiaofan]

Even a free tier Oracle Cloud server located in US is not too bad.

ubuntu@ubuntuopc1:~$ ping -c 6 apicn.hualaikeji.com
PING apicn.hualaikeji.com (120.92.43.182) 56(84) bytes of data.
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=1 ttl=50 time=160 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=2 ttl=50 time=160 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=3 ttl=50 time=160 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=4 ttl=50 time=160 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=5 ttl=50 time=160 ms
64 bytes from 120.92.43.182 (120.92.43.182): icmp_seq=6 ttl=50 time=160 ms

--- apicn.hualaikeji.com ping statistics ---
6 packets transmitted, 6 received, 0% packet loss, time 5214ms
rtt min/avg/max/mdev = 159.753/159.888/160.142/0.128 ms
ubuntu@ubuntuopc1:~$ ping -c 6 s.taobao.com
PING v6-sh-sz-zb.wagbridge.alibaba.taobao.com.gds.alibabadns.com (106.11.94.18) 56(84) bytes of data.
64 bytes from 106.11.94.18 (106.11.94.18): icmp_seq=1 ttl=31 time=153 ms
64 bytes from 106.11.94.18 (106.11.94.18): icmp_seq=2 ttl=31 time=153 ms
64 bytes from 106.11.94.18 (106.11.94.18): icmp_seq=3 ttl=31 time=153 ms
64 bytes from 106.11.94.18 (106.11.94.18): icmp_seq=4 ttl=31 time=153 ms
64 bytes from 106.11.94.18 (106.11.94.18): icmp_seq=5 ttl=31 time=153 ms
64 bytes from 106.11.94.18 (106.11.94.18): icmp_seq=6 ttl=31 time=153 ms

--- v6-sh-sz-zb.wagbridge.alibaba.taobao.com.gds.alibabadns.com ping statistics ---
6 packets transmitted, 6 received, 0% packet loss, time 5006ms
rtt min/avg/max/mdev = 152.742/152.881/153.099/0.121 ms

 

[xiaofan]

I'm using ViewQwest and having China websites issue as well. My WeChat is just cannot connect to the network at all most of the time in the past 3 days.

When Wechat was disconnected, Steam was also unable to connect to the Friends network...

For Taobao, the website front page can be partially loaded, some resources are failed to load. If you do search in Taobao, will take a while to load or cannot load at all.

I thought it was an issue from ViewQwest, now when I see Singtel user is also facing same issue, I feel better...

Try changing DNS to see if that helps. It is said that Viewquest DNS server has some issues.

Vuewquest uses static IP address, not so sure if you can try asking them to change the IP for you or not, this may help as well.

 

[miloaisdino]

Okay the ONR is limited.

But why SingTel can not clear DHCP leases? What kind of backward DHCP server are they using?

they scared from the old incident where singtel was fined a $500k for dhcp server outage. the dhcp server was consistently 90% cpu (lol) and running out of ram (prob using pagefile), and when someone attempted to update software.. well rip

https://www.zdnet.com/article/singtel-slapped-with-fine-for-broadband-service-outage/

 

[xiaofan]

they scared from the old incident where singtel was fined a $500k for dhcp server outage. the dhcp server was consistently 90% cpu (lol) and running out of ram (prob using pagefile), and when someone attempted to update software.. well rip

https://www.zdnet.com/article/singtel-slapped-with-fine-for-broadband-service-outage/

Yes I remember this incident caused by DHCP server and the previous incident caused by fire.

Hopefully their infrastructure is more stable now. Indeed I do not remember any incident after this one .

 

[SeanPC]

I restart the ont and router. Now is ok.

 

[xiaofan]

I restart the ont and router. Now is ok.

Great.

Did you check if your IP address changed?

If yes, then it is probably related to the particular IP address. (Most likely case)

If not, it could be related to the router.

 

[SeanPC]

Great.

Did you check if your IP address changed?

If yes, then it is probably related to the particular IP address. (Most likely case)

If not, it could be related to the router.

I not sure lol