In reply to a post by radar:

In reply to a post by Ixel:

I see.

I'm waiting for an update to my ticket, though it won't be till later today I think.

Here's a speed test I did last night in the early hours of the morning: https://www.thinkbroadband.com/_assets/speedtest/but...

Here's a speed test I did just a minute ago: https://www.thinkbroadband.com/_assets/speedtest/but...

It has to be congestion. My ping graph doesn't look impressive either - http://i.imgur.com/TBrMlPW.png

Zen have finally fixed my packet loss and poor single thread speed problems.
https://www.thinkbroadband.com/speedtest/15051995351...

So lets hope it stays this way.

Dave

Nice! Mine is still an ongoing issue unfortunately, but there is an underground line fault apparently so I'm waiting for that to be resolved (causing SNRM swings periodically as well as a few other issues such as abnormally high FEC errors) though DLM is doing a job of masking the fault to some degree at the moment with INP 3 and delay 8ms. So, as Zen have told me, once this is fixed then they'll look into the packet loss/single threaded speed issue I'm having too. The results are worst in the evenings I've found, until a little after midnight. Sometimes the multi threaded speed test gives me 10-15 megabits during the evening even though my connection isn't under heavy use. I've currently got a Ubiquiti EdgeRouter ERPro-8, a very powerful router that can process over 2 million packets per second and around 8 gigabits per second I believe. It's connected to a HG612.

TBB speed test I did earlier - https://www.thinkbroadband.com/_assets/speedtest/but...

Buffer bloat isn't an issue too - https://i.imgur.com/2ZtKOLB.png

Pretty bad as you can see.

I'm expecting to hear tomorrow about what happened today with the underground engineer (who was one day overdue), as something happened this morning which increased my line attenuation by quite a fair amount but the underlying issues are still there.

Edited by Ixel (Thu 14-Sep-17 22:22:08)

Just to post an update on my situation.

The underground engineer fixed whatever fault there was however there's still a problem with an abnormally large amount of FEC errors, and more CRC errors recently which made DLM change me to INP 4 delay 12ms from INP 3 delay 8ms. I've not noticed significant fluctuations in SNRM though for all I know the INP/delay could be masking that as it's currently masking the continuous CRC errors I had.

Subsequently a phone/voice engineer turned up but was unable to help me as his equipment showed everything passed. He suspects I might be on a faulty port at the DSLAM but it's difficult to verify that without risking further call out charges. The only potential way I can verify that is if I order another line which is what I've now done. Family is still occasionally telling me that there's no dial tone on the line every so often but if there's still an intermittent fault then it's going to be difficult to prove this. I'm not getting re-syncs with the DSL side though. The second line is being installed mid October by which time I should know more about whether it's potentially a faulty DSLAM port or how the second line performs compared to the current line. Credit to Zen however as BT tried to charge me for the voice engineer not finding a fault but I was told there was no agreement about potential charges this time so I'm off the hook and it'll be disputed with BT .

Currently I'm waiting for a follow up regarding my packet loss/slow single threaded download speed, last I heard I'm being moved to another peer or something.

Final update to say the single threaded download speed issue has been resolved, I imagine my packet loss issue will disappear too but I need time to determine that. Thanks to Stefan and Andrew at Zen for helping resolve the problem!

https://www.thinkbroadband.com/_assets/speedtest/but... - Nice single threaded download speed now

Test was run over IPv6, its worth just checking using the Run IPv4 button, usually v4 and v6 will take same routes but always a chance of different paths

I see. Well I did the test again on IPv4.

https://www.thinkbroadband.com/_assets/speedtest/but...

Looks fine I believe, I know it slowed down slightly a little later on during the test but I'm not the only one browsing the internet here right now so I think that's why.

EDIT 2: Speed tester anomaly. I did a 1GB file a few times and consistently getting around 50 megabits per second according to the EdgeRouter and based on Chrome's MB/sec speed.

Edited by Ixel (Fri 29-Sep-17 23:17:53)

Oh dear, looks like problems are returning again.

https://imgur.com/a/3XPww

Anyone else having issues, any thoughts?

EDIT: Swinging SNR is back too, see https://imgur.com/a/ojlPM

Edited by Ixel (Wed 04-Oct-17 21:23:14)

Hi,

Yes, I am still having these issues as well.



I was told last Friday that Zen's network operations team were taking a look, but haven't heard anything since.

It took a week or so before I heard back regarding what the network operations team said or the fact they've fixed it.

This morning my speeds are reasonable again, but I'm half expecting the problem to return again. Then there's the unresolved and apparent intermittent line fault. Second line is still scheduled to be installed on the 13th going via the TT Wholesale back-haul so can't wait to see how it compares to my current Zen line.

Also seening the same persistent single thread issues that get progressively worse as peak 8-10pm approaches.

Lines sync is 71.4Mbps/ 20.0Mbps fastpath (eci cab) open profile.



John and the network team implemented a remote change on the ZyXEL equipment in the local exchange, which provided a temporary fix. The issue is reappearing again.

Could this be a traffic policer (shaper) issue or somthing like a virtual switch issue, that talktalk wholesale were recently struggling with?

Edited by Nigel_FL (Thu 05-Oct-17 20:04:18)

I see. Yeah you could be right. It does feel like traffic's being shaped at times, whether by accident or intentionally I don't know.



Just did this a moment ago, nobody in the house is using a lot of the potential throughput either.

Edited by Ixel (Thu 05-Oct-17 20:35:11)