Unifi Connection Issue to International Server, Occured usually at night 9-11pm
|
|
 Jul 1 2021, 08:41 PM
Return to original view | Post
#21
|
 
Senior Member
1,748 posts Joined: Mar 2010 
|
QUOTE(heLL_bOy @ Jul 1 2021, 08:31 PM) you can check till now.. CDN77 latency is quite stable and also they using PCCW same as M247 But look at the time in his two graphs. The exact same time CDN77 improved and M247 deteriorated. Too much of a coincidence to put blame solely on M247, no?is m247 endpoint back to TM got issue.  |
|
|
|
|
|
Jul 1 2021, 09:10 PM
Return to original view | Post
#22
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(heLL_bOy @ Jul 1 2021, 08:53 PM) source to destination both are using the PCCW same pipe and if really got issue both also will have the same problem. This is what my [mtr --aslookup sg5-wireguard.mullvad.net] says. Educate me please, bro;destination back to source might be not the same routed back. CDN77 is using Equinix SG back to TM for M247 i not sure what pipe routed back maybe you could check at your backend. CODE ThinkPad (192.168.1.138) 2021-07-01T21:01:35+0800 Keys: Help Display mode Restart statistics Order of fields quit Packets Pings Host Loss% Snt Last Avg Best Wrst StDev 1. AS??? stupid.router.me 0.0% 72 2.0 6.1 1.2 107.1 17.5 2. AS4788 60.53.xx.xxx 0.0% 72 4.1 6.6 3.0 41.1 7.2 3. AS??? 10.55.106.5 0.0% 72 10.3 16.0 9.0 126.1 20.1 4. AS??? 10.55.50.52 0.0% 71 8.8 10.7 8.0 43.2 5.2 5. AS??? 10.55.37.90 0.0% 71 9.6 12.5 8.5 119.4 14.0 6. AS3491 63-218-166-1.static.pccwglobal.net 0.0% 71 12.1 13.9 10.5 80.3 10.6 7. AS3491 HundredGE0-5-0-0.br03.sin02.pccwbtn.net 42.3% 71 30.8 37.5 30.8 150.1 20.2 8. AS3491 HundredGE0-5-0-0.br03.sin02.pccwbtn.net 41.4% 71 31.7 35.8 30.6 124.6 15.2 9. AS3491 198148metro.tenGigE0-6-0-20.10.br03.sin02.pccwbtn.net 39.4% 71 35.9 35.7 26.0 73.8 8.8 10. AS??? 37.120.220.219 40.0% 71 33.7 57.1 33.7 145.6 27.9 11. AS9009 193.27.15.40 32.4% 71 34.7 33.9 26.5 78.6 7.4 12. AS9009 94.198.43.50 38.0% 71 37.1 34.2 25.7 149.9 18.6 This post has been edited by G-17: Jul 1 2021, 09:11 PM |
|
|
Jul 1 2021, 09:17 PM
Return to original view | Post
#23
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
I think maybe @hostingmalaya should try to sell a copy of his aduan/report as a NFT for 0.002435887543 BTC, because I feel this problem won't ever go away.
 |
|
|
Jul 1 2021, 09:40 PM
Return to original view | Post
#24
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(michaelkkl @ Jul 1 2021, 09:25 PM) This is mtr from TM to m247 SG. Okay, but the problem now is cannot even connect back to the same SG server. Even when managed to connect to a different server (sg7 vs sg5 previously) and mtr back to my IP but the trace doesn't even complete.Hop no.6 is PCCW KUL and pass to PCCW SG on hop no.7 and finally reaching m247 in SG. You may connect to your m247 vpn and try mtr back to your tm 60.53 ip to see how the return route look like. CODE My traceroute [v0.93] ThinkPad (10.104.205.96) 2021-07-01T21:36:14+0800 Keys: Help Display mode Restart statistics Order of fields quit Packets Pings Host Loss% Snt Last Avg Best Wrst StDev 1. AS??? 10.64.0.1 26.9% 108 31.0 32.5 30.9 66.7 4.1 2. AS9009 89.38.225.49 25.0% 108 32.2 32.1 30.9 34.8 0.8 3. AS9009 193.27.15.41 24.1% 108 40.0 38.5 31.7 85.1 9.2 4. AS??? 37.120.220.218 25.0% 108 33.8 33.0 31.1 44.1 2.3 5. AS??? 27.111.229.206 48.1% 108 51.1 51.5 48.1 73.9 4.1 6. (waiting for reply) more than 100 packets sent and still (waiting for reply) at hop #6. Is it perhaps some iptables rule or my router's firewall preventing the trace? O_o? Fuck it lah. Tonight just use JP server. This post has been edited by G-17: Jul 1 2021, 09:45 PM |
|
|
Jul 1 2021, 09:54 PM
Return to original view | Post
#25
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(heLL_bOy @ Jul 1 2021, 09:48 PM) base on your MTR.. CDN77 also having same route path back to source from hop 5 (27.111.229.206) Turns out I entered the wrong IP. here's the full readout;so its M247 issue. CODE My traceroute [v0.93] ThinkPad (10.104.205.96) 2021-07-01T21:49:36+0800 Keys: Help Display mode Restart statistics Order of fields quit Packets Pings Host Loss% Snt Last Avg Best Wrst StDev 1. AS??? 10.64.0.1 29.4% 51 30.1 28.2 27.1 30.1 0.7 2. AS9009 89.38.225.49 30.0% 51 28.1 30.3 27.7 53.2 5.4 3. AS9009 193.27.15.41 38.0% 50 31.8 37.1 29.3 90.3 13.4 4. AS??? 37.120.220.218 34.7% 50 28.6 29.3 27.0 34.4 1.8 5. AS??? 27.111.229.206 59.2% 50 48.9 48.5 47.5 52.4 1.3 6. AS4788 60.53.xx.xxx 73.5% 50 56.5 57.1 56.0 58.6 0.9 So is there anything I can tell Mullvad? Last time they already switched some transit partner when I had issues, now need to ask them again. But then, once TM switch some other routing next week then problem start again how?.... Sigh, it was soooo damn good the last couple of Level3 days, then the TM idiots went back to PCCW junk and this started again. This post has been edited by G-17: Jul 1 2021, 09:59 PM |
|
|
Jul 1 2021, 10:17 PM
Return to original view | Post
#26
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(heLL_bOy @ Jul 1 2021, 10:03 PM) this is from CDN77.. you can see both using 27.111.229.206 back to TM Nah, I don't trust majority of commercial VPN providers. Basically any VPN that ask for email during registration and don't accept Monero, I won't use. For now I'll use Mullvad's JP servers. Latency is higher a bit, but no packet loss and relatively high download speeds at the moment.CODE HOST: 10g-singapore-kvm-1-89-187- Loss% Snt Last Avg Best Wrst StDev 1. AS60068 89.187.162.60 0.0% 10 0.1 0.1 0.1 0.1 0.0 2. AS60068 185.229.188.170 0.0% 10 0.1 0.1 0.1 0.2 0.0 3. AS??? 27.111.229.206 10.0% 10 17.3 17.1 16.9 17.3 0.0 4. AS??? ??? 100.0 10 0.0 0.0 0.0 0.0 0.0 5. AS4788 58.26.239.30 0.0% 10 9.2 9.2 9.1 9.3 0.0 or you consider to have another backup vpn other then this? previously maxis fiber also having the same issue like now its turn out is m247 issue itself. Previously had my own WireGuard tunnel running on a DigOceanSG Ubuntu droplet, but that was more for performance than privacy, and I shut it down when this Unifi problem got unbearable for me last month. Sign, I'm going to collect all this problematic AS numbers and buy Magnum when lockdown ends. This post has been edited by G-17: Jul 1 2021, 10:19 PM |
|
|
|
|
|
Jul 6 2021, 08:30 PM
Return to original view | Post
#27
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
Tried switching IP pool. Curiously, I ended up with some wierd tar-xxx-xx.tm.net.my hostname. Never seen that before. Some university pool? Anyway, still no luck. Connection to M247-SG is dogshit.
CODE ------------------ Mullvad Not Connected: mtr -r -w -c10 sg5-wireguard.mullvad.net ---------------------- Start: 2021-07-06T18:29:41+0800 HOST: ThinkPad Loss% Snt Last Avg Best Wrst StDev 1.|-- router.asus.com 0.0% 10 4.0 2.0 1.2 4.0 0.8 2.|-- tar-204-254.tm.net.my 0.0% 10 4.0 4.4 3.2 7.0 1.1 3.|-- 10.55.106.7 0.0% 10 11.1 10.6 10.0 11.8 0.7 4.|-- 10.55.100.191 0.0% 10 11.0 10.8 9.2 15.5 1.8 5.|-- 10.55.37.84 0.0% 10 14.4 12.1 11.5 14.4 0.9 6.|-- 63-218-166-1.static.pccwglobal.net 0.0% 10 13.3 13.3 12.2 14.9 0.9 7.|-- HundredGE0-5-0-0.br03.sin02.pccwbtn.net 60.0% 10 34.0 33.5 33.2 34.0 0.4 8.|-- HundredGE0-5-0-0.br03.sin02.pccwbtn.net 50.0% 10 33.4 33.4 32.6 34.3 0.6 9.|-- metro.tenGigE0-6-0-20.10.br03.sin02.pccwbtn.net 60.0% 10 33.0 33.1 32.7 33.6 0.4 10.|-- 37.120.220.219 60.0% 10 33.8 43.2 32.1 73.0 19.9 11.|-- 193.27.15.40 40.0% 10 29.8 28.3 27.3 29.8 0.9 12.|-- 94.198.43.50 60.0% 10 27.6 27.8 27.5 28.1 0.3 ---------------------- Mullvad Connected: mtr -r -w -c10 tar-204-254.tm.net.my -------------------------- Start: 2021-07-06T18:32:02+0800 HOST: ThinkPad Loss% Snt Last Avg Best Wrst StDev 1.|-- 10.64.0.1 30.0% 10 29.4 27.0 26.3 29.4 1.1 2.|-- 89.38.225.49 50.0% 10 27.6 27.1 26.5 27.6 0.4 3.|-- 193.27.15.41 20.0% 10 30.2 40.0 27.3 66.6 14.6 4.|-- 37.120.220.218 30.0% 10 27.4 27.6 27.0 29.1 0.7 5.|-- 6939.sgw.equinix.com 60.0% 10 42.9 31.4 27.0 42.9 7.7 6.|-- 65.49.109.190 30.0% 10 48.9 49.5 48.9 50.4 0.5 7.|-- tar-204-254.tm.net.my 90.0% 10 50.4 50.4 50.4 50.4 0.0 |
|
|
Jul 7 2021, 03:48 PM
Return to original view | Post
#28
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
Been strongly considering switching to Maxis Fiber. Does anyone know how the switch works. According to this page, I walk into TM Point, ask to transfer service and pay outstanding Unifi bill. After that need to wait for confirmation. But do I register with Maxis first before doing this, or after? Most important is that I don't want to experience many days with no internet while waiting for the switch.
|
|
|
Jul 9 2021, 10:13 PM
Return to original view | Post
#29
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(hostingmalaya @ Jul 9 2021, 09:52 PM) You guys still having issue to Singapore Server ? Yes for me. for a couple of commercial VPN providers. Those that employ M247 servers have been bad for days. Basically packet loss almost whole day ever since that day they switched from Level3 back to PCCW routing. Since yesterday VPNs that use CDN77 also having bad (~40%) packet loss and frequent disconnections. Actually, it's even worse now compared to last month before TM's so called fix. For the commercial VPN provider I trust the most, I'm basically unable to use their SG servers since July started. Currently posting this from JP server now.On the rented VPS side, I only have experience with DO. Fired up a new droplet yesterday to run a WireGuard+DNSCrypt tunnel, and so far it's been okay. Latency could be better (last year I could get below 10ms), but overall it's been stable with no packet loss. Aside from SG servers, I even got Youtube buffering yesterday and today on my Android TV, tested on both local and international content. A lot of FaceTime video dropped calls this past week also, and I double checked with mobile data (Celcom) and calls didn't drop anymore, so must be TM issue. TL;DR - been a sh!t week overall. This post has been edited by G-17: Jul 9 2021, 10:30 PM |
|
|
Jul 11 2021, 03:58 PM
Return to original view | Post
#30
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(muok @ Jul 11 2021, 11:39 AM) M247 is back to normal! Yay. Don't celebrate lah. Now you went and jinxed it already, surely will koyak again tomorrow.  This post has been edited by G-17: Jul 11 2021, 04:00 PM |
|
|
Jul 21 2021, 11:11 PM
Return to original view | Post
#31
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
Last 10 days seems to be good for me connecting to DigitalOcean or Mullvad (M247) Singapore servers. Average ping is usually between 15ms (DO) to 25ms (M247), though sometimes will spike. Packet loss has been minimal (like 0.0X% per 24 hrs). Crossing fingers this will continue. Will post update in a couple of weeks or if/when got problem again.
. SG-Mullvad over WireGuard tested with Viewquest server. UniFi500  This post has been edited by G-17: Jul 21 2021, 11:13 PM |
|
|
Jul 28 2021, 11:19 PM
Return to original view | Post
#32
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
Last hour or so also suddenly had higher latency with my DO tunnel. Still usable, but feels more like a US server latency instead of the country next door.
|
|
|
Aug 13 2021, 09:40 PM
Return to original view | Post
#33
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
My DigitalOcean-SG tunnel last few days been getting latency spikes (sometimes reaching 100ms+) around 9pm to midnight.
Anyone else? I assume it's the same Equinix issues? This post has been edited by G-17: Aug 13 2021, 09:42 PM |
|
|
|
|
|
Aug 19 2021, 02:29 PM
Return to original view | Post
#34
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
Rebate?... By TM??.... In Malaysia???
Sure or not?? Now the poor crony is RM15 poorer. So kesian. If we got $RM0.10 for ever packet we lost, we all can buy Richard Mille like our songlap politicians. In other news, yesterday evening and today morning M247 koyak again.  |
|
|
Aug 26 2021, 01:08 PM
Return to original view | Post
#35
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
TM decided to f*ck up the SG-M247 route again. Unusable since around 12 noon.
Sigh, at least DO-SG is okay now. |
|
|
Aug 26 2021, 07:55 PM
Return to original view | Post
#36
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(muok @ Aug 26 2021, 07:45 PM) I think it's an M247 issue this time and not a TM one. Now seems to be okay though. Did you do that trace now or earlier when the problem occurred (roughly 12 - 3 pm)? This is from Tokyo to M247 SG. Funny, for me the JP servers were okay (latency of course higher, but 0% packet loss) at the time SG servers koyak. Now SG, HK, JP all seems to be working like normal. Haven't 9pm yet. Last few days the latency spike starts around that time, so who knows.... This post has been edited by G-17: Aug 26 2021, 07:55 PM |
|
|
Aug 26 2021, 08:41 PM
Return to original view | Post
#37
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(muok @ Aug 26 2021, 07:57 PM) The trace was done just now. This is from my Unifi connection. Seems the route from my home to Mulvad's JP-M247 servers is different from yours. Don't know if it's to do with my home's IP pool (currently 60.52.xxx.xx) or Mullvad's peering/transit setup with M247-JP. My trace to all Mullvad's WireGuard Tokyo servers seem to go through Level3. Total 10 hops. 8 hops if I minus my router and home IP. I'm on the phone now and it's a bit inconvenient to paste the report from the app I'm using, so excuse my lack of trace paste. QUOTE(go626201 @ Aug 26 2021, 08:19 PM) http://smokeping.mywebping.com/smokeping/?...N.VPN-NordVPN-5 Your NordVPN-SG-M247 graph seems to be similar to MullvadVPN-SG-1. But I notice my Mullvad connections usually seem to mirror your graph for MullvadVPN-SG-2. I don't really know what's the difference between the two and what the peering/routing differences are. Crap TM... Now i just use my ovh sg to port redirect vpn service. So far so good. My connection uses Mullvad's WireGuard servers: sg4/sg5/sg6/sg7/sg8-wireguard.mullvad.net For me over the last few days (not including the problem today afternoon), the ping spike and packet loss started around 9pm to around 11pm or 12midnight, similar to your SG-2 graph, I think. This post has been edited by G-17: Aug 26 2021, 08:48 PM muok liked this post
|
|
|
Aug 27 2021, 05:26 PM
Return to original view | Post
#38
|
|
Senior Member
1,748 posts Joined: Mar 2010
|
QUOTE(heLL_bOy @ Aug 27 2021, 04:14 PM) different ip prefix will get different upstream to certain destination. not all destination reaching are under same upstream becos of balancing. So in other words, I better hope there's no lightning strike that trips the power and reboots my router to a different prefix range. muok ip traceroute via NTT yours ip traceroute via Level3  This post has been edited by G-17: Aug 27 2021, 05:36 PM |
| Change to: |  0.0243sec
 0.54
 7 queries
 GZIP Disabled
Time is now: 26th November 2025 - 01:36 PM |
Quote