50M : Post RS Errors

[Sephiroth]

Originally Posted by Kushan

You would be much better off calling them, they can check for issues there and then, as well as book an engineer if need be.

Good luck with that if you get the script monkeys!

[Jon22]

Which is exactly why I haven't

[Kushan]

Which is why your issue is still ongoing. Even the "script monkeys" are (usually) capable of checking the outage page and booking a tech. There's a reasonable chance that it would have been fixed by now had you just called in a couple of days ago.

[Sephiroth]

Go on. Do it and let us know how the conversation ran! It's a percentage game. Good luck!

[Jon22]

Ha, I would but the support team have replied back on the forums to book an engineer. In less than 24 hours as well.

[Sephiroth]

Originally Posted by Jon22

Ha, I would but the support team have replied back on the forums to book an engineer. In less than 24 hours as well.

Ha ha - Kush!

[Jon22]

Noticed a novel thing with the recording of errors by the SH2ac. Once it gets to 200000000, it goes in to negative numbers and counts upwards.

Downstream
DS-1 DS-2 DS-3 DS-4 DS-5 DS-6 DS-7 DS-8
Frequency (Hz) 299000000 267000000 275000000 283000000 291000000 307000000 315000000 323000000
Lock Status(QAM Lock/FEC Sync/MPEG Lock) Locked Locked Locked Locked Locked Locked Locked Locked
Channel ID 13 9 10 11 12 14 15 16
Modulation 256QAM 256QAM 256QAM 256QAM 256QAM 256QAM 256QAM 256QAM
Symbol Rate (Msym/sec) 6.952000 6.952000 6.952000 6.952000 6.952000 6.952000 6.952000 6.952000
Interleave Depth I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17
Power Level (dBmV) 4.32 4.10 4.09 4.35 4.37 4.59 4.52 4.68
RxMER (dB) 32.50 28.00 28.94 29.81 31.07 33.72 34.63 34.63
Pre RS Errors 5140313 -1494310897 1385274806 528430914 69359201 255807 51686 55772
Post RS Errors 7827 15968 9034 8199 9790 10211 12189 13226

[Sephiroth]

That's professional software development for you. Influenced by someone who said "Nah there will never be so many RS errors. The phenomenon you've seen is an effect known as "overflow" leading to the "negative complement"; at least the code was written so that the SH didn't crash.

Your SNR is appalling.

[qasdfdsaq]

200,000,000 is a weird number to overflow at. Could it perhaps have been 2,000,000,000 (actually 2,147,483,647)?

[Edit]
Probably was 32-bit signed int overflow. I can see a reading of 1,385,274,806 in there which is over 200,000,000 already. He probably missed out a '0'. 2 billion errors is a lot to have on one channel, but considering routers these days are rated in mpps (million packets per second) it wouldn't take that long at full pelt.

[Jon22]

Oops yeah sorry missed off a 0. Seem to be getting around 100,000 errors after every refresh (think it refreshes around every 5 seconds if you keep hitting the refresh button)

[Kushan]

Originally Posted by Sephiroth

Ha ha - Kush!

I don't see why that's amusing, the point still stands - phoning in, even to offshore, is 9 times out of 10 the fastest way to get an issue resolved. The forum support team is good, but it still takes time and if everyone went via the forum, it'd take days to spot local issues rather than hours.

As inept as some offshore agents are, even the worst are capable of booking a tec and had OP just phoned in, he'd have likely had a tech the next day. Now this has gone on for several days.

Originally Posted by qasdfdsaq

200,000,000 is a weird number to overflow at. Could it perhaps have been 2,000,000,000 (actually 2,147,483,647)?

[Edit]
Probably was 32-bit signed int overflow. I can see a reading of 1,385,274,806 in there which is over 200,000,000 already. He probably missed out a '0'. 2 billion errors is a lot to have on one channel, but considering routers these days are rated in mpps (million packets per second) it wouldn't take that long at full pelt.

My first thought as well. A surprisingly common occurrence, these days. Then again, I can understand why the developer never gave much thought to the data type, a signed 32bit integer should be enough for anyone :P

[Jon22]

Understand what your saying Kushan, only reason I haven't is that it doesn't seem to be causing a noticeable effect on the connection. I still get full speed and the ping monitor looks ok:



I'm quite happy to wait for an engineer. Having said all that, I've just had another look at the downstream stats and the SNR has gone back to what it was. I've no idea whether I actually need an engineer now or not?

DS-1 DS-2 DS-3 DS-4 DS-5 DS-6 DS-7 DS-8
Frequency (Hz) 299000000 267000000 275000000 283000000 291000000 307000000 315000000 323000000
Lock Status(QAM Lock/FEC Sync/MPEG Lock) Locked Locked Locked Locked Locked Locked Locked Locked
Channel ID 13 9 10 11 12 14 15 16
Modulation 256QAM 256QAM 256QAM 256QAM 256QAM 256QAM 256QAM 256QAM
Symbol Rate (Msym/sec) 6.952000 6.952000 6.952000 6.952000 6.952000 6.952000 6.952000 6.952000
Interleave Depth I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17 I=12
J=17
Power Level (dBmV) 3.14 3.16 3.10 3.32 3.36 3.47 3.49 3.73
RxMER (dB) 34.48 34.35 34.35 34.48 34.63 34.63 34.35 34.35
Pre RS Errors 14472186 1700150773 -1460133266 1116306445 164902362 2184339 1003479 1397140
Post RS Errors 28762 2164659 222446 25030 39024 44539 51789 55981

[Sephiroth]

It all depends on the Post-RS errors and whether or not they are rising. With those SNR levels, it should not be rising much if at all, but the Pre-RS count would be rising significantly; they don't matter if Post-RS is not rising.

It's power level and SNR stability that's required here and an engineer is required as first step if there is no stability. Not that he can do much other than check all the physical bits and your cabinet tap. Then he can escalate to Networks.

[Jon22]

The Post RS errors are still going up even though the SNR hasn't changed from the last post. Engineer is booked for Thursday, no idea what (s)he'll be able to do, probably say everything is fine

[jb66]

Probably pretend the hub is faulty