IPFreely
Fibre optic
Joined 3 years ago
User Profile
User Widgets
Contributions
Re: Another Hub 5 user with latency issues...
Upstream channel 1 isn't happy. Your downstream power levels are out of kilter as well: shouldn't be an over 2 dBmV variation between neighbouring channels as there is between 5 and 6. Your upstream channel layout looks like you have fibre outside getting converted into coax but your downstream SNR is a bit lower than I'd expect and your upstream transmit power way higher. QoS will clearly fix the RF/optical path.7Views0likes0CommentsRe: NEW PACKAGE - upstream channel missing
Slightly different topic to keep focus no evidence of overselling from your BQM you posted. The profile on the 3.1 upstream channel changing is for the same reasons as the 3.0 channels lowering modulation. It's to try and cope with noise. The reduction in the 3.1 channel was a widespread thing, not just you, to improve reliability. The channel was notched as there was a narrow band of interference. High power radio transmissions get onto the cable network as unterminated coax points in homes act like antennae.2Views0likes0CommentsRe: NEW PACKAGE - upstream channel missing
If that was your postcode in a screenshot you posted earlier I can see why the service isn't great where you are. That kind of property is a nightmare to manage in terms of noise ingress which can in turn break everything for 250-500 premises. Very much depends on the people living there as much as anything else and can be literally nothing the cable company can do about it if people aren't cooperative.5Views0likes0CommentsRe: High latency errors (full logs & BQM attached)
Useful. The latency isn't worse at peak times, no capacity issue evident. There is a clear SNR issue, the 3.1 upstream's modulation is flapping... 25-04-2025 13:51:59 notice US profile assignment change. US Chan ID: 6 25-04-2025 14:02:02 notice US profile assignment change. US Chan ID: 6 25-04-2025 14:10:36 notice US profile assignment change. US Chan ID: 6 25-04-2025 15:12:47 notice US profile assignment change. US Chan ID: 6 ...which will cause latency spikes. There are clearly uncorrectable FEC errors on it so there'll be retransmits, causing latency spikes, and a level of packet loss if it gets too high. Channel Channel Type Number of Active Subcarriers First Active Subcarrier (Hz) T3 Timeouts 6 OFDMA 200 74000000 59 The 3.0 upstreams all have T3 timeouts. The network as a whole has a high noise floor. You can tell this by looking at the lowest upstream channel which sits in the noisiest part of the spectrum. Channel Frequency (Hz) Power (dBmV) Symbol Rate (ksps) Modulation Channel ID 4 23600000 45.3 1280 QAM 64 5 This would normally be 6.4 MHz wide taking it down to 20.4 MHz, it's been adjusted to 1.6 MHz wide leaving it between 22.8 and 24.4 MHz to avoid the noise it would be hitting between 20.4 MHz and 22.8 MHz. This comes with a loss of capacity of 75% of the channel however this was presumably preferably to having it constantly dropping modulation and even then erroring. There are likely long-standing and difficult to resolve issues on the local network. VM engineers will try and patch it up and get them all running smoothly but in the longer term the fix is Project Mustang. Messing with QoS will do absolutely nothing the issue is at layer 1. Took me less than a minute to read the original post and less than 10 to compose this one. Being conscientious isn't a big time sink, far better than posting random irrelevant content on QoS/BQM wasting the time of everyone reading and your own writing it.8Views1like0CommentsRe: High Latency and Packet loss for gaming and websites.
Odynson90 wrote: Smaller MTU sizes allow for smaller packets of data to be sent quickly. That isn't how it works, dude. It's a maximum not a target. Small packets are small because they've a beginning and an end and not much in between: when they're complete they're sent full MTU, or in case of WoW MSS as it uses TCP, or not. WoW stuff used to risk being delayed by Nagle's Algorithm but that was turned off when the 'optimize network for speed' was added. Your PC will have a 1500 MTU set ensuring the 2000 isn't an issue. The PC will not try and build anything larger than 1500 bytes. You likely are able to send with a 1472 payload size on the ping whether Hub MTU is 1500 or 2000: this is standard. If you had been able to capture a session between you and WoW before it would have had an MSS of at most 1460. 1500-40 byte TCP/IP header. Unclear what was up there, might be a software bug in the Hub but the 2000 MTU doesn't impact you as nothing you're sending has an MTU above 1500. Many other people haven't touched the MTU on their Hub and are fine pointing away from that as the cause but I would need to test and I'm certainly not getting a Nexfibre service connected for that 😁17Views0likes0CommentsRe: Latency expectations
How far north are you going? Unfortunately you connect to a Broadband Network Gateway and are probably 'hard wired' to go to the site you're currently at. There might be a better route to get you to the BNG but VM's internal routing isn't great on the whole as it follows the cable network rather than taking more direct paths.83Views0likes3Comments