Because by tech, only static IP is meaningful to customer, lol.

Anyways, still testing the cloudflare, still need to bet luck at sliver thing to get fulll speed, else just 100kb/s download + 80ms and 80% chance get connection hang, lol.

I tested Firefox http2, not change, still depend on that sliver thing, and it will fallback to H2 when detect too high packet loss.
At least it is CloudFlare loadbalance issues, so just wait and see how la.

This post has been edited by BenYeeHua: Aug 29 2025, 10:32 PM

Last time I terminate my acc by going to TM point, u guess what?
It still going on for 2 months, and charged me extra RM 10 for their fault, which don't cut off within 30 days.

I already confirm nothing else should be done(including receive sms/call) from that TM point(mitc), yet still kena lol.
---
So far there is nothing for LOS issues, the signal stay the same during raining, guess Seremban just upgrade the fiber then.
Only left is the old CloudFlare issues la, SG bad and KUL random bad after 8 PM.

This post has been edited by BenYeeHua: Aug 30 2025, 02:16 PM

I guess it depend on area, at my Uni area, the house got pulled 2-4 fiber, because port a lot, lol.
Even 1 ppl pulling via the Windows, yes, WINDOWS.

Wow, so much hang now...

Welcome to the country where 8.8.8.8 can get bottleneak lol.

Including other links to SG, HK etc.
But funny, speedtest to fast and digi malaysia are fine, both <15ms, guess they are high priority connection.
---
While IPv4 lagging, v6 fine, lol.



This post has been edited by BenYeeHua: Sep 26 2025, 11:50 PM

Not DNS issues, it is the IPv4 for my area bursted.
You can see the bottleneck is route #2 which is 100.89.191.254.

Ya, yesterday night strange, only v4 bursted, v6 running like nothing happen, might be something happen at the CGNAT.

---
For DNS, it still depend.
Like Steam, using Google DNS bring me to server without throttled by TM, while 1.1.1.1 doing the reverse.

But, Google DNS kick me to HK IPv6 Aliyun, while 1.1.1.1 bring me to Malaysia IPv4 Aliyun(by not return any IPv6 address).
A lot of times Google DNS doing this, a far far away IPv6 server, so...

It is still kind of depend, like yesterday night, sure kick me to IPv6 solved the network issues.

This post has been edited by BenYeeHua: Sep 27 2025, 09:18 AM

packet loss are same on my side, but yesterday night was, you wanna 90% packet loss at IPv4, or 30-40% packet loss at IPv6, bad or worst only 2 choice.

Using Google DNS DoH(chrome)
chrome://net-internals/?#dns

Resolved IP addresses of "h5api.m.taobao.com": ["2401:b180:7003::3e0","59.82.121.14"].
No alternative endpoints.

Resolved IP addresses of "h5api.m.taobao.com": ["59.82.121.14"].

Resolved IP addresses of "h5api.m.taobao.com": ["2408:4001:f10::2dd","47.246.165.34"].

3 times(wait for TTL timeout), different results.

Can be it asking by rotate in between 8888 8844(both IPv6) la.
----
First can understand, it is return oversea for both IPv6 and v4, and it is located at Zhe Jiang China(IPv6) and BeiJing China(IPv4)...
Second is strange, IPv6 missing.
Third one is more strange, BeiJing(IPv6) and SG(IPv4)

I was observed this strange F5 slowness when grabbing Red Packet here, so found out this IPv6 issues, but I never found out IPv4 also can be in China, until now.
https://huodong.taobao.com/wow/z/tbhome/pc-growth/tao-coin?
----
FYI, CloudFlare DoH is

Resolved IP addresses of "h5api.m.taobao.com": ["47.246.165.34"].

So if the only server is SG, going to CloudFlare solved the most, and it will not give you strange oversea IPv6 address.
But when server is MY or SG, like Lazada and Shopee, it is a headache to choose in between Google DNS and CloudFlare DNS...
(Most of the time both platform works, just sometimes it required me to turn off and on my phone WiFi, or cut off at chrome://net-internals/?#sockets by flush sockets pools, solved.)

Like steam, using Google DNS give full speed to upgrade Steam Beta client, while CloudFlare DNS have high chance getting slow speed during Night.
Guess it is solved on Steam side, as both giving same SG server now.

As above, unsure why Google DNS prefer to jumping around.
This is also why I default to CloudFlare DNS, it might not giving the best Malaysia server as result, but it will ensure it is SG server.

WoW, you still using AC1200v2?
It is always a bet that it gonna last 30 days uptime, or <24 hours uptime.

But yes, the NAT Boost is better than orange router la, now the skyworth router not bad without Nat Boost, I guess the processor higher raw power solved it.

And yes, you can ask for new router, as they will prefer to upgrade your modem for higher speed stability, but it will be Modem + Router combo, if you wanna more stable, then you need to buy another router.
But, based on me tested for years, this skyworth combo router/modem good enough liao.

Even if they don't give a new one, you can ask for replacement and agree to resign 24 months, lol.

This post has been edited by BenYeeHua: Sep 27 2025, 11:32 PM

1. Possible, that explain why BiliBili which used Aliyun before also gaming Google DNS with IPv6, but only at US.
(It gone after not using AliYun)

2. Anycast only perform for certain IP address, they don't like that because how's China network works.
Like I said before, for Malaysia user, anycast IPv4 in China just redirect you via US, then China Telecom.(As TM default to that)
Having split IP address helps their complex China network work well, the quota is expensive, specially cross border.

Even transfer in between 3 big ISP causing a lot of issues, they had to setup 3 line for 3 ISP, and pay at least 10x of price for each line...
20 years ago they also split into 3 game server for 3 ISP, now just pull 3 line for servers la, also any Steam gamer in China just use booster(proxy) to redirect P2P traffic at server center, and only same Tencent oversea line available for all proxy.
Also lucky, China's video platform disabled PCDN for oversea customer, else it gonna be hell to watch any video over P2P, lol.

3. Yes, old news liao.
TM never useful to those China traffic, 20 years ago all line to China can get 50ms(just ensure it is 60.50.x.x high QoS), 10 years ago I can get 50ms by connect to China Mobile, now, going to HK also bottleneck liao.

Nope, 2 years with GN630V, online at least 30 days.



You should check the GPON uptime, then PPPoE uptime.

If GPON uptime > PPPoE uptime a lot, then having another router might solved it(If it is not on PPPoE server side issues).
If router uptime > GPON uptime, call tech to check the fiber.

Yes, based on discussion here, the only problem is hardly to get both technical works together, either TM come and Maxis no come, or reverse.

This post has been edited by BenYeeHua: Sep 29 2025, 09:16 AM

Yes, China is Business pay the price, while others is Customer pay the price, with some interesting like SG and Thailand is cheap Internet as solution.(SG I know, center is there, unsure Thailand etc)
And also HK already slowly taken control by 3 ISP in China, they provided a few cheap internet and getting famous, just it will be bad news for us, as it will means less traffic for oversea traffic.

As I recall got new cable incoming, just checked.
https://www.submarinenetworks.com/en/systems/intra-asia/alc
https://www.submarinecablemap.com/submarine...-link-cable-alc
RFS
2027 Q4

Should enough?
---Off-Topic, maybe I shared already? Really no memory liao lol.

---
Anyways, back to real life experience, as long as I kick my Bilibili live stream video from m3u back to flv, there is stable connection available in Malaysia, the problem is the API call overhead(for m3u) is too high.
d1--ov-gotcha05b.bilivideo.com and d1--ov-gotcha05.bilivideo.com stable(can get <1s browser buffer), unstable when IPv4 unstable.
d1--ov-gotcha07b.bilivideo.com and d1--ov-gotcha07.bilivideo.com sometimes unstable(need buffer >2s), but got IPv6.

For m3u, only d1--ov-gotcha208b.bilivideo.com stable, quick check show it is TM and zenlayer MY.
Other like 207 207b, yuck.



---
Another news for today, 3 China ISP got license to launch and operate Traffic/Internet satellite, so if lucky, Asia can get hand for sea customer to use the satellite, when 6G is ready?

This will greatly improve internet coverage in Asia, specially for forest country like us, only problem is price.

Did you try disable EasyMesh and WPA3?

So far, EasyMesh causing trouble when hotspot enabled nearby, and it will background scan causing spike, just like Windows.
I recommend this also, if your windows device got issues, but for Intel, set the roaming to restrain when high/good RSSI in driver settings enough liao, only MTK need this.
https://www.reddit.com/r/BoosteroidCommunit...wlan_optimizer/
https://www.martin-majowski.de/
(But latest Intel got removed, lol)
https://community.intel.com/t5/Wireless/Ple...ng/td-p/1623942
Android no need, all latest Android 12+ already enabled throttling background scan in dev settings.

Yes, EashMesh will perform another background scan, specially for hotspot, I turn on an unsetup smart socket and it caused my 2.4Ghz stop working.
EashMesh also not following the radar scan interval setup that you set in WiFi 5Ghz, so spike gonna happen from time to time.


WPA3 is no internet, yes.
Because the device connected by asking WPA3, but sometimes the router suddenly stop accepting WPA3 and only allow WPA2, your device will stop connecting as it is not running as WPA3.
After I set to WPA2 only, all my Android device asking me to forget the network or refuse to connect, as it disallow WPA2 handshake.

This post has been edited by BenYeeHua: Sep 29 2025, 09:25 PM

Yes, agree you at tech point of view, the money customer saved for subs any ISP at lower cost, not just bring them basic access of internet, but also bring them to the evil part of who gonna bear the cost.
And well, it is also them, as hidden cost.

Now they just stuck with most video bitrate around 4Mbps, PCDN getting them ban for uploading anything to cloud etc, even live streamer also get such issues and had to sign up for RM 800 plans for just pure upload without throttling.
Even if PCDN implement well, it can save cost by having mostly local traffic, but nope, all they want is to sell those server/service/content provider expensive quota, and they only dare to ban ISP's customer to "solve" the issues.
(Another reason is they no need to pull a lot of local fiber, this one I am kind of unsure on tech part la, like how many it gonna take to fulfill the requirement for PCDN.)

And not just that, 1 hour ago, I get block again, because I am oversea IP.
https://zh.moegirl.org.cn/Mainpage
https://img.moegirl.org.cn/common/thumb/f/f...s_2_Treaser.jpg

The orig pic is fine, just thumb over IPv6, lol.

Still, any place has it own network issues la, like TM, 10 years ago, they dare to throttle any ADSL to Google traffic as low as 480kbps, including Google search, and it only solved after fiber online....
That time, the only thing you can hope for YouTube full speed, is you get kick/chosen by Google to connect TM peer server, and DNS give you TM peer IP.

Just take a try first, should be something like radar detection and WPA3.
If same, then it means router issues.

And back for complaints, lol.

1. zenlayer at Malaysia broken, Tencent side working.
https://static.hdslb.com/js/jquery.min.js

If connect to Zenlayer like 2602:ffe4:c20:110::10d or 162.128.167.237, failed to loaded.
If connect to tencent MY like 43.174.2.195, success

First case broken in MY, last like few weeks.
Should be something broken from MY to HK/CN, at least for Zenlayer.

Edit: check back another browser, after timeout, I get this message, lol.

So yup, Zenlayar to HK/CN.

---
2. After power cut off and reboot, IPv4 routing from 175.141.17x.x to cloudflare MY broken, like others
IPv4/v6 having 80ms latency to CF MY, but IPv6 can reach higher speed without high packet loss, v4 can't.


Funny, download overloaded and only got 100-1024 kbps, while upload 20-50M lol.

For HoYoVerse game, they use CloudFlare for unknown reason, using special CF IP 104.18.13.147, all HoYo game broken at night, stop login because failed to connect or timeout.
Yesterday testing around 11 PM, found out random jumping at TM side, including router #2- router #N not responding to traceroute.

Guess issues on TM side, this time?
Gonna test a few more days, might try reboot router to jump back 115.134.x.x if still the same.
---
3. Amazon MY having issues after power and reboot, unsure SG affected or not, gonna test later 0 AM.

Amazon MY which is sg-public-api.hoyolab.com IP address 54.192.117.45 etc failed to respond on time, while it is holding the previous request, it can still respond to another next few request lol.

As the main server is at SG, already write host file to connect directly to Amazon SG, gonna see it works at 0 AM or not.
---
4. For Tencent, QQ server already slow upload/download for image at night, last for 1 month liao.
But not that CloudFlare slow, still got 0.1-10 Mbps la, lol.

So far that's it, nothing break too much, until you use those imfamous website, or CloudFlare website lol.

---
Oh ya, forget, the WARP to KUL and Johor is getting broken a lot of times, it is because the connection/routing keep getting changed, so it hang the connection until you reconnect to WARP again.
But funny enough, Johor still got lower latency 10ms and higher speed test/full speed, lol.



This post has been edited by BenYeeHua: Oct 19 2025, 10:06 PM

So far unifi 99% uptime for me la, the only problem is only power cut off few days ago, then back online lol.

After power back online, can see inside the router, the fiber got signal/reading, but no TX packet received, lol.
Surprised the mobile line still running while power off.

Or, when I reboot router, and get shared IP address with those bots, and caused blacklist at server side.
Still, public IP is a should thing to get for UniFi users, as there is lot of bots running around and MCMC don't care about it.

And again today, they send cloudflare-dns.com to blackhole lol.
Now all using cloudflare-dns.com as DoH will keep getting DNS error because heavy packet loss.