Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[broadstairs]

I did mention this in the G.INP and ECI thread but it has got worse.

Prior to G.INP being enabled on my line I was interleaved and saw the usual FEC errors downstream but my upstream was pretty quiet and had very very few errors of any sort apart from the odd burst.

Now since I had G.INP enabled downstream only the other morning I have seen huge (and I mean huge) numbers of UPSTREAM FEC errors usually during the peak evening hours, where there were very low levels before. Now this morning I can see that I have had about ~60000/min FECs continuously since about 19:40 yesterday evening. My router is a ZyXEL VMG8924-B10A running f/w version 11 if that has any effect on things!

What I dont understand is why my upstream now seems so noisy when in the past it was quite quiet and the vast majority of errors happened downstream. You can see my stats on MDWS as user broadstair (note no S on the end!).

Stuart

[Ronski]

At work we used to regularly get 600,000 FEC on the down stream (can't remember what the US used to do), for weeks on end, then it would just disappear. Perhaps its g.inp related or just coincidence for you.

I'm still waiting for g.inp both at home and at work , I'm running a ZyXel at home so we'll see what happens when mine goes live, thought mine might have done by now seeing as we're not too far apart.

[broadstairs]

At work we used to regularly get 600,000 FEC on the down stream (can't remember what the US used to do), for weeks on end, then it would just disappear. Perhaps its g.inp related or just coincidence for you.

I'm still waiting for g.inp both at home and at work , I'm running a ZyXel at home so we'll see what happens when mine goes live, thought mine might have done by now seeing as we're not too far apart.

Yes it will be interesting to see another locally. As I said these errors are upstream which is something which has ONLY started to appear since my G.INP was enabled.

Stuart

[broadstairs]

I decided to take my router down and update it to V13 of the f/w (was V11), left for 30mins and back now. So we'll see if the errors being reported are still there.

Stuart

Edit: My FECs upstream have now dropped to zero so either the ZyXEL needs a reboot when G.INP is enabled or there might be a f/w bug. ANyway I'll see how it goes as the day(s) progress.

[broadstairs]

Well after about 24 hours since the reboot I am still seeing errors upstream during the evening. Now what I dont understand is why G.INP being activated downstream it has caused my upstream to start showing errors which did not happen pre G.INP. It makes no sense. Now I know everyone says FECs dont affect DLM but none the less any errors of this magnitude has to have some kind of performance impact and why on earth should the downstream G.INP affect upstream errors?

Stuart

[Chrysalis]

FECs will have no performance impact but just work your modem (and dslam) harder to correct them.

Did none of the FEC related settings on your upstream get changed when g.inp got enabled?

[broadstairs]

FECs will have no performance impact but just work your modem (and dslam) harder to correct them.

Did none of the FEC related settings on your upstream get changed when g.inp got enabled?

Well if the router and dslam have to work harder then that in itself will have a performance impact.

Not sure where to look or what settings to look at.

Stuart

[Chrysalis]

only if cpu is saturated which shouldnt be the case as these devices are designed to operate in such modes.

When I say work harder I mean slightly harder not significantly harder.

[ejs]

The sending end won't be working any harder, the FEC data is generated in advance of when it is used, so it will be doing the same amount of work regardless of if the FEC data ends up being used at the receiving end or not. The devices at each end will have dedicated hardware for doing things like the FEC process.

[broadstairs]

OK so lets assume no particular performance issues....

So why when pre G.INP my upstream was virtually error free do I now see during the evening peak times hundreds of thousands of FECs over a period of maybe a couple of hours. To me it makes no sense at all to suddenly see all these errors. Something has changed but how do I find out what and why?

Stuart

[ejs]

Do your stats have DSLAM/MSAN type: IFTN:0xb204 or IFTN:0xb206 ? There is a mention of "Increased Upstream FEC Count" in the list of fixes in a ECI NGA release notes document.

[underzone]

Do your stats have DSLAM/MSAN type: IFTN:0xb204 or IFTN:0xb206 ? There is a mention of "Increased Upstream FEC Count" in the list of fixes in a ECI NGA release notes document.

Good spot 

[broadstairs]

So yes my DSLAM is IFTN 0xb204. So basically there is a bug in the DSLAM f/w which causes increased FECs upstream. My FECs started to increase tonight again at around 21:00 to tens of thousands per minute and has continued about every minute since.

I suspect a fix for this will be as rare as the proverbial rocking horse droppings!

Stuart

[NewtronStar]

So yes my DSLAM is IFTN 0xb204. So basically there is a bug in the DSLAM f/w which causes increased FECs upstream. My FECs started to increase tonight again at around 21:00 to tens of thousands per minute and has continued about every minute since.

I suspect a fix for this will be as rare as the proverbial rocking horse droppings!

Stuart

Could it be they need a new G.INP fix Mk3  for ECI

[WWWombat]

Do your stats have DSLAM/MSAN type: IFTN:0xb204 or IFTN:0xb206 ? There is a mention of "Increased Upstream FEC Count" in the list of fixes in a ECI NGA release notes document.

Good spot 

Good spot indeed, but it raises an interesting question...

Is that fault one which induces real FEC faults/corrections? Or is it a fault that causes a bad count to be reported?