Just setup a wireguard vpn on Linode SG. CPU power not enough or speed should be higher...

Confirm had improvement today, but still have about 30% need to be fix...

Better hope TM dont do any changes to DO.
Now got some ip prefix from DO is connection friendly to unifi.
And always good.
So if later TM change here change there and causing all DO prefix boom together then GG.com ..haha

The TM always have 2 different routing to DO with different DO ip prefix.
So just rebuilt instances and get the always stable IP prefix is enough...

Fresh boom...

Normal issue,always like this one.

Last year,my solution for family is get a singapore cloud server/vps,and set port forward to the corporate vpn ip:port.
And insert the vpn ip as the server ip.

To be clear the problem already long long time,not a new issue. (Happens every certain timeframe.)

This post has been edited by go626201: Jun 3 2021, 09:42 PM

Can u try to traceroute the ip,or try reboot the router to get another block of unifi ip prefix.

Yes,but can confirm latency of the last 2 node which should be the gateway of the singnet network.

Not seeing on my end. all normal at this current time.

Normally few days or few weeks will happens once.
I think it is just random happens only,not deliberately by Unifi.
Just sudden disconnect only.
Now Unifi network overloaded,so i think it is quite normal to have some problem.

HE.net to Unifi also not stable one...
Last time ovh to unifi via HE.net also alot of problem....

Latency to HKG always 40+ms so this is normal... (Unifi only)



(Unifi via Wireguard VPN-Linode-SG)


This post has been edited by go626201: Jun 11 2021, 02:17 AM

Telekom Malaysia Server Speedtest


Singapore
ViewQwest


Starhub


MyRepublic


M1


NewMedia Express


OVHCloud SG


The reason why u getting 100ms++ is because u are using ipv6.
Unifi to OVH ipv6 always will get 100ms++ latency. so that is normal.


This post has been edited by go626201: Jun 11 2021, 02:12 PM

Better complain to both. TM only ensure the line is usable but optimize is not their 1st priority ,so u need to complain to CloudFlare to let them contact TM to solve the problem.

Actually u can get the another ip range that different from the speedtest.
I think 3 cloud server company got different peering pipe with unifi.
Like Linode,u can get 40ms ip address and 10ms ip address(Via Equinix SG). (For 10ms one,currently 100% stable connection between Linode SG and Unifi on day and night) (I currently have the Linode SG 10ms network instances so far so good.)
Digital ocean is also same concept,40/100ms ip address and 10ms ip address (Via Equinix SG).
Vultr i dont remember.(I know they got their own Choopa peering line)







Seems like yesterday had a spikes. But overall is good


This post has been edited by go626201: Jun 12 2021, 05:28 PM

Past 10 days,got spikes but the latency only add within 5ms only so it is normal. (To be clear the packet lost is still might be a problem,but only 0.1% of the whole time might having packet loss,so it is acceptable)


Those cloud server didn't have contract so if got problem then just terminate and change to another host,so Linode is okay for use now.

And for second problem.The answer is yes,different unifi ip prefix might get better connectivity to specific location.
But for recent problem,i think all unifi ip prefix is having the problem,but some might be better.(But still crap la!)
A simple example for using different unifi ips prefix.
I using mi smart sound box(CN version),but some ip prefix can get 75% stability to use the sound box and enjoy music. (This also might have QQ music problem,but better)
But if get the worst ip prefix,the box might unable to get respond from the server and QQ music might cant work properly(maybe 1 song can 1 song cant,or buffering issue)
QQmusic have too many cache server,so sometimes to 1st ip good,but 10mins later connect to another ip then no good.

Unifi ip pools should not have difference from plan,i think the IPs u can get is most depends on your area.
(But sometimes might get a rare ip prefix,
mostly is a list that u can get easily like
60.48 60.51
115.135 175.137
147.158 (I prefer this pool,seems like a new ips by unifi in 2019)
Any others ips that i not wrote here is because i never get it.(Or i only get once per year,so i dont remember)
More ips can see this link
https://forum.lowyat.net/topic/4878047

So this really depends on your usage.
And different unifi ip prefix might get abit higher or abit lower latency then the previous unifi ip prefix. (About 3-4ms+-)

Overall,i suggest dont care the ip prefix just get multiple vpn,and see which one is usable for that current time is better solution now.
Until TM spend money to fix the problem,i dont think this can be solve anysoon.
Currently TM is doing this crap way " tear down the east wall to repair the west wall".
And this is not the first time for having this kind of issue,but the severity for this time is too too too high ,how come TM can having this issue for almost a month...

If the problem like affect 1% of the usage then i can agree TM just dont care about it...
But this problem affect almost 60% of normal usage for long time,then this i dont know what to say... (Cloudflare Fastly etc all popular service also affected how come like this...)

This post has been edited by go626201: Jun 12 2021, 07:03 PM

maxis only using tm infrastructure for last-mile connectivity to provide service to users.
the bandwidth after local exchange center will go through maxis own network.

Hi guys,i had changed my smokeping nat reverse proxy to Cloudflare.
So the link had permanently change to
http://smokeping.mywebping.com/smokeping/
Sorry for the past few hours of interruption.
Anyone can now use the link to report to unifi or something else without changing link.