Kitz Forum
Broadband Related => FTTC and FTTP Issues => Topic started by: 111888 on May 01, 2022, 09:26:42 AM
-
Its been 5 days of uptime, I have on the upstream 508 errors. Downstream 0 errors.
Is this in an acceptable range?
-
That rather depends what type of errors you are referring to. If it's Errored Seconds (ES) then it sounds just fine.
A screenshot of the page showing the errors would assist.
-
I've moved this question to the more appropriate forum section, and also deleted the duplicate of the same question.
-
here is attachment
I have been having some issues with playing online games, where latency is important. latency in game is showing its fine, but it seems everyone is like 2 secs ahead of me, could be the servers.
-
What are the reported latencies in game?
When you 'ping' websites in Windows, what kind of latencies do you see?
It looks like you are using a TP-Link router, based on your screenshot. It looks like the errors you can see are CRC errors. If this number isn't increasing all the time, it's not worth worrying about.
-
it is increasing all the time. those errors its now at upstream error: 531
every day upstream error is increasing, even during the day.
In game latency is fine. Below 30% latency. pinging websites shows 8ms ping.
-
I would only worry about severely errored seconds. (SES).
SES occurs when is too many CRC at once.
Errors are normal on a DSL line.
-
Very low error rates really.
-
Not sure what its reporting seeing as it says Pkts. I'd assume its ES though, I have 341/day and rock-solid stable.
Line Uptime: 134d 17h 21m 13s
Annex: B
Line Mode: G.993.2 (VDSL2)
Profile: 17a
G.INP: D: Not Enabled / U: Not Enabled
Attain Data Rate: 86.765 Mb/s / 24.371 Mb/s
Actual Data Rate: 79.995 Mb/s / 20.000 Mb/s
Impulse Noise Prot: 0.0 sym / 0.0 sym
Interleave Delay: 0.0 ms / 0.0 ms
NFEC: 255 / 255
RFEC: 16 / 16
LSYMB: 21451 / 5410
Interleave Depth: 1 / 1
Interleave Block: 255 / 255
LPATH: 0 / 0
Line Attenuation: 11.8dB / 13.4dB
Signal Attenuation: 11.8dB / 13.2dB
Noise Margin: 4.1dB / 8.0dB
Transmit power: 14.0dBm / 3.3dBm
FECS: 1599837 / 2681385327
ES: 10466 / 45697
SES: 0 / 17
LOSS: 6 / 0
UAS: 145 / 145
HEC: 0 / 0
CRC_P: 0 / 0
CRCP_P: 0 / 0
15m Code Violations: 3 / 4
15m FEC Errors: 26 / 17
1d Code Violations: 115 / 140
1d FEC Errors: 3243 / 2113
Unfortunately a good connection doesn't guarantee good online gaming, as some games penalise people on stable connections to offset those who are not, so you end up with a worse experience.
-
As mentioned above, TPlink modems record packet errors which are CRCs. The count will continue increase as long as the modem is up. It is errored seconds that are the important ones as far as DLM is concerned. 508 CRCs in 5 days is negligible.
>> but it seems everyone is like 2 secs ahead of me
You may be able to get some more stats using StatPOSTer (https://forum.kitz.co.uk/index.php?topic=14377.msg269624#msg269624) which works on most TPlink modems.
[Moderator edited to fix typo.]
-
Its easy to forget overhead lines will pickup some radio frequencies and underground may pass other cabling. They've marked outside my DP now and the cable seems to pass direct across the electrical feed to the houses.
Even someone parking their car above the ducting with a bad alternator can cause a sudden burst of errors.
Worth also noting though how much better my line is now since they replaced a damaged drop wire yet from the line stats a problem wasn't entirely obvious until one line failed entirely.
-
Have you tried using a different router yet?
Regarding the CRC errors, if these aren't increasing every second, I doubt it would make much difference.
-
I would agree. From my experience of experimenting with Fritzbox and Zyxel, I found no difference. I therefore don't believe that matching the chipset with the Exchange is crucial (or if it is, it has minimal impact on the improvement of the quality of the connection), as modern Lantiq chipsets are as good with Broadcom-based cabinets.
-
500 CRCs in 5 days is nothing, realtively speaking
That's 500 errored packets, a packet is between 70 bytes and 1.5Mb.
Anytime these errors are captured, the packet is resent by the source
You'll have no issues in the real world with these kinds of errors
-
I would agree. From my experience of experimenting with Fritzbox and Zyxel, I found no difference. I therefore don't believe that matching the chipset with the Exchange is crucial (or if it is, it has minimal impact on the improvement of the quality of the connection), as modern Lantiq chipsets are as good with Broadcom-based cabinets.
On a short line with high sync the chipset probably doesn't make a significant difference. However, on longer lines I can assure you it can very much make the difference between a stable, low errors, connection and an unstable ,high errors, one.
I say this from the experience of trying many different routers and chipsets on my ~800m line. From those I have tried a BCM63168 works best on my line. The common mode filter in my Zyxel VMG8924-B10A also gives an improvement on models without one.
-
500 CRCs in 5 days is nothing, realtively speaking
That's 500 errored packets, a packet is between 70 bytes and 1.5Mb.
Anytime these errors are captured, the packet is resent by the source
You'll have no issues in the real world with these kinds of errors
500 CRC can be 1 ES or 500 ES.
Whether any data is resent or not depends entirely on what the user is doing at the time (TCP/UDP/etc).
It could be that nothing is resent.
They are on different layers so there's not necessarily a correlation between the 2.
-
On a short line with high sync the chipset probably doesn't make a significant difference. However, on longer lines I can assure you it can very much make the difference between a stable, low errors, connection and an unstable ,high errors, one.
I say this from the experience of trying many different routers and chipsets on my ~800m line. From those I have tried a BCM63168 works best on my line. The common mode filter in my Zyxel VMG8924-B10A also gives an improvement on models without one.
Plus on an ECI cabinet with a really clean line, using a Lantiq modem that lets you tweak the SNRm is useful.
I only get full sync thanks to pushing a 4dB SNRm due to increased crosstalk over the years.
-
Plus on an ECI cabinet with a really clean line, using a Lantiq modem that lets you tweak the SNRm is useful.
Same benefit for Huawei cabinets. Not all lines are deemed worthy of a sub 6dB target by the DLM.
-
Same benefit for Huawei cabinets. Not all lines are deemed worthy of a sub 6dB target by the DLM.
There are lines stable enough that wont just trigger DLM to increase the SNRm?
-
Same benefit for Huawei cabinets. Not all lines are deemed worthy of a sub 6dB target by the DLM.
I have done it three times, and all three times I got banded. :) So at least in my experience the SNRM tweaking you have to be careful as the DLM doesnt seem to like you having a SNRM way below the target. I guess ok if its not extreme though, like 5db instead of 6db. I had a feeling it would happen this third time but given my line because of the FEC/interleaving cant sync above 74 anyway at 6db I gave it another go. So didnt lose anything from the banding this time.
-
I have done it three times, and all three times I got banded. :) So at least in my experience the SNRM tweaking you have to be careful as the DLM doesnt seem to like you having a SNRM way below the target. I guess ok if its not extreme though, like 5db instead of 6db. I had a feeling it would happen this third time but given my line because of the FEC/interleaving cant sync above 74 anyway at 6db I gave it another go. So didnt lose anything from the banding this time.
That's what I would expect as clearly DLM has a reason for not letting you have 3dB whereas on ECI its simply because there is no lower setting.
-
That's what I would expect as clearly DLM has a reason for not letting you have 3dB whereas on ECI its simply because there is no lower setting.
The DLM only seems to apply banding to ECI lines for high ES.
I've never seen it apply banding on a Huawei line for high ES.
G.INP means even when pushing a line to 3dB the ES only increases a little especially when compared to the hundreds you get on fastpath.
-
Just for clarification ES was pretty much zero in my case, I did it on an interleaved line.