Connection issues after joining - packet-loss, latency spikes, poor performance

Forums › Quic › Connection issues after joining - packet-loss, latency spikes, poor performance

RedeviL

21 posts

Geek

#312848 23-May-2024 11:33

Hey all,

Joined Quic on a Sprinter plan (untagged DHCP) on Monday 20th - what a fantastic quick and easy migration I might add!
Prior to leaving Orcon I was not having any issues which would suggest a LFC fault.

Ever since joining I have been experiencing a consistent ~10% packet-loss to all internet destinations.
Powering off the ONT and on again clears the packet-loss for roughly 5 hours, after which it returns and stays.

With and without the PL - downstream flows cause quite significant jumps in jitter.
This is most easily seen and reproduceable with Speedtests: https://www.speedtest.net/result/16276338229
Upstream performance is less than expected at all times.

The following is what I have tried:

Investigated and excluded the internal network - no PL observed between wired or wireless clients.
No PL between multiple wired and wireless clients following different paths between the APs, switch and router.
Tested ping to multiple internet destinations (eg: 103.243.102.32, 1.1.1.1, 8.8.8.8) all exhibit loss but at different times.
Running a MTR to 1.1.1.1
Speedtest results slower than expected and latency during download fluctuating significantly (12-250+ms)
Testing ping from an external connection to my Quic connections IP, observed same PL %.
Power cycled ONT, Router and Switch.
Cable re-seated for all connections.
Alternative cable between Router and ONT tested.

MTR:

I have submitted a Fault, as I am quite stumped.
But while waiting for Quic to get in touch I thought it sensible to see if anyone else has any suggestions?

My setup:
TP-Link Deco BE65 in Wi-Fi Router mode - IPv6 Enabled
TP-Link Deco BE65 in meshed AP mode
Hard wire backhaul at 2.5Gbps through NetGear MS510TX Switch

1 | 2 | 3

meow
13243 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#3233744 23-May-2024 11:38

The fact this is happening directly after your router is pretty interesting. I wonder if switching to PPPoE (even to test) will help at all? Problem is, it takes around ~30mins to switch to PPPoE as you need to wait for your DHCP lease to fully expire.

So here is mine - you'll note we're on the same BNG also with no loss. I'm using Untagged DHCP also:

Start: 2024-05-23T11:37:26+1200
HOST: smokepi Loss% Snt Last Avg Best Wrst StDev
1.|-- 10.0.2.1 0.0% 10 0.2 0.2 0.2 0.3 0.0
2.|-- bng1-akl1.vetta.net 0.0% 10 11.9 11.6 10.8 12.1 0.5
3.|-- xe1-2100-146.pe2-akl1.vet 0.0% 10 11.2 11.6 10.8 12.3 0.6
4.|-- et21-00-401.core1-akl1.ve 0.0% 10 11.8 11.4 10.9 11.8 0.4
5.|-- as13335.akl.ix.nz 0.0% 10 12.6 12.5 11.7 13.2 0.5
6.|-- one.one.one.one 0.0% 10 13.3 12.0 11.1 13.3 0.7

Will be interested to see the outcome here!

Michael Murphy | https://murfy.nz
Referral Links: Quic Broadband (use R122101E7CV7Q for free setup)

Are you happy with what you get from Geekzone? Please consider supporting us by subscribing.
Opinions are my own and not the views of my employer.

RedeviL

21 posts

Geek

#3233746 23-May-2024 11:50

Yeah, as soon as it leaves the router it is affected by what ever is wrong.

That is actually a great idea about PPPoE - I do need to head out for ~30 mins so can change and it should be online when I return.
Will report back then!

RedeviL

21 posts

Geek

#3234097 24-May-2024 10:31

Update:
PPPoE made no difference overall. Same performance, same jitter.
It went from ~10% PL to ~20%, turned out the weird TP-Link config was having IPv6 recreate a PPPoE session.
So IPv4 was flapping up and down ever 20 seconds.

After fixing that it fell to a consistent ~1%.
Ran that for a couple of hours, no improvement just slower overall connection responsiveness expected from PPPoE.

So switched back to DHCP.

Talking with the Quic guys now - has everyone confused and scratching their heads.
They're going to test in their lab.

I am actually starting to wonder if it's a weird provisioning state on the ONT.

HoochWindgrass

11 posts

Geek

#3234666 25-May-2024 19:15

I signed up just to join into this post, but as someone who just swapped over i'm having similar issues from Auckland.

Lots of latency, packet loss, slow uploads and downloads.
I've just come from BigPipe and had a solid 900+/450+ connection, but now struggle to hit 700/200....That with the packet loss and ping times are making me worried I've picked and ISP that Auckland's not ready for (the better speeds I've seen listed seem to be out of Auckland people - from Chch or Wellington)

It appears there are routing issues, and I never get p/l internally, its all once it hits routers outside my network.

I'm at a loss what else I can try either.

michaelmurfy

meow
13243 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#3234667 25-May-2024 19:35

Coming from BigPipe confirm your MTU is 1500. Know plenty who are even on Hyperfibre without issues so it’ll get fixed either way.

HoochWindgrass

11 posts

Geek

#3234669 25-May-2024 19:52

Thanks for the tip, they were not defined, which I assumed meant they would be 1500 - bad assumption, I guess...
I changed them and now seem to be hitting 900+/500+ which is great. Packet loss seems better too.
I'll keep my eyes on it but that seems to be a lot better already. Thank you!

*** Edit - I spoke too soon, its back to being poor with speeds/packet loss.
It was looking better for a very short time...

RedeviL

21 posts

Geek

#3234763 26-May-2024 12:10

Welcome HoochWindgrass!

A FYI - I found switching from DHCP to PPPoE yielded a more stable connection.
Not running correctly, but at least consistently usable.

Update on my findings:
Both Friday and Saturday mornings the connection was unusable - continually flapping up and down.

You can see the ~10% PL originally experienced, followed by the periods of none (after a power cycle of the ONT).
The spike on the 23rd is when I tried PPPoE for the first time. This spike was TP-Links PPPoE IPv6 being weird.
Followed by the small spike returning to DHCP.

Friday I tried powering off the ONT and Router for 10 minutes.

This worked until about 8PM at which point I had to do it again.

Saturday morning I had more time, so I tried multiple angles.

I powered off the Router for 5 minutes - returned worked for a minute then went back to flapping
I powered off the Router for 10 minutes - returned worked for a minute then went back to flapping
I power cycled the ONT - returned worked for a minute then went back to flapping
I powered off the ONT for 5 minutes - returned worked for a minute then went back to flapping
I powered off the ONT for 10 minutes - returned worked and has been stable

I ran pings and traceroutes at each step - when it was flapping I got zero response after the router.

Just no replies - the only time I got a Network Unreachable from the router was when the ONT was off.

I discovered I can extract the system logs from the TP-Link Decos web UI - ran through them and correlated the times.
Nothing stands out, it would come online - get a DHCP lease and come online.

The only thing I saw was when it starts flapping the device logs a lot of:

Sat May 25 10:10:55 2024 daemon.err conn-indicator: [send_query] Failed to send DNS query: Network unreachable

I believe the issue lies with the ONT, I don't think it is a hardware or fibre fault as it was working flawlessly prior.

I believe it is the port Provisioning.

Have asked Quic if they can re-provision the circuit with Chorus.

Sharesies kids

Free kids accounts - trade shares and funds (NZ, US) with Sharesies (affiliate link).

HoochWindgrass

11 posts

Geek

#3234767 26-May-2024 12:21

I appreciate your efforts to find the cause!

Which method did you use to initiate the conversation with them? - I guess Ill have to try the same.

My static IP lists my location as being in Canterbury not Auckland which I suppose doesn't help - might not even be relevant.

That said, I'll try switching back to PPPoE and see if it's any better. That's not really a 'fix' though but if it helps its worth it

RedeviL

21 posts

Geek

#3234768 26-May-2024 12:32

I logged into the Account section of the Quic website.
Clicked on the Service.
Then clicked Lodge Fault under LFC Integration.

But yeah, PPPoE is no solution - simply a means to use the connection reliably while the issue is looked into.

HoochWindgrass

11 posts

Geek

#3234771 26-May-2024 12:38

Thanks, ill try that method.

Here's todays results

HoochWindgrass

11 posts

Geek

#3234772 26-May-2024 12:42

See this bit worries me - it makes me reluctant to reach out

RedeviL

21 posts

Geek

#3234777 26-May-2024 13:15

That's if it goes to the LFC.
They come to physically investigate and find the fault is at your end - then the fee is applicable.

Quic reviewed the details and got in touch - so the LFC part isn't applicable.

I also logged a Problem under Support.

Mikek

108 posts

Master Geek

#3234778 26-May-2024 13:18

HoochWindgrass:

Thanks, ill try that method.

Here's todays results

So whats the issues here with this one ?

HoochWindgrass

11 posts

Geek

#3234781 26-May-2024 13:32

Nothing really, just showing it for detail compared to the previous.
My primary issue now is the inconsistent speeds I get from various websites/domains.
I typically had very stable download rates in my speedtest logs for example, where now it's very inconstant and fluctuates a lot.

You can see where i switched on the 24th

michaelmurfy

meow
13243 posts

Uber Geek

Moderator

ID Verified

Trusted

Lifetime subscriber

#3234782 26-May-2024 13:43

HoochWindgrass: My primary issue now is the inconsistent speeds I get from various websites/domains.

Just FYI - that isn't the norm:

One thing I do though is rate limit my upstream on my router so I don't hit Chorus's rate limiters which can be quite aggressive past 500Mbit.

Do recommend logging a fault though. The thing I've noticed with people having problems is they're all on BNG2-AKL2 but is only affecting some people suggesting there is perhaps an underlying bug here.

1 | 2 | 3