GPROTECH's profile

New Contributor

 • 

2 Messages

Monday, August 24th, 2026 9:48 PM

Business circuit failing daily - case escalated twice, repair date is 9 days out

Looking for help getting a technician dispatched sooner than the date we have been given.

Case number [Edited: "Personal Information"]. Business Internet 2 Gigabit at a commercial site. Happy to
provide the account number and circuit details privately.

The problem
-----------
Traffic beyond our modem disappears for seconds to minutes at a time, several times a
day, while the modem itself stays perfectly healthy. We monitor every device on the path
once per second, continuously, and have done since 17 August.

Today alone we recorded six separate outages, the worst losing 79.4% of packets. Last
Friday it reached 100% loss twice. The trend is getting worse, not better - 1 degraded
minute on the 18th, 3 on the 19th, 5 on the 20th, 27 on the 21st.

Why I do not think this is our equipment
----------------------------------------
In every one of those minutes:

 - Our modem answered every single probe at 1-3 ms, its normal speed
 - Our router, four switches and four VLAN gateways were all at 0.00% loss
 - 20% to 100% of packets past the modem were lost

We also run identical monitoring at a second Comcast Business location, which shares the
Comcast network from the same aggregation point onward. During every episode that second
site measured 0.00% loss to the same Comcast routers, at the same moment, while
simultaneously measuring this site at 25-100% loss.

Two sites, same infrastructure, one failing.

What has already been tried
---------------------------
The modem was restarted remotely on 21 August. The fault returned 32 minutes later and
was back to 88% loss within the hour. The restart also cleared the modem's error
counters, so that day's diagnostic history was lost.

Separately, the circuit is provisioned at 300 Mbps upstream and we have never measured
better than about 83 Mbps.

What I am asking for
--------------------
1. A technician dispatched before 31 August.

2. That whoever attends reads signal levels and error counters BEFORE resetting or
   replacing anything - upstream and downstream power and SNR at both the modem and the
   tap, plus T3/T4 timeout and uncorrectable codeword counts. Every reset so far has
   erased the evidence we need.

3. Confirmation of whether the CMTS logged errors for this modem at our recorded
   timestamps, and whether the upstream is actually delivering its provisioned speed.

I have a one-page technical summary with per-minute measurements ready for whoever
attends, and can provide raw per-second logs on request. Please let me know how to get
the account details to you privately.

Oldest First
Selected Oldest First

Official Employee

 • 

49 Messages

2 days ago

It is very nice to meet you here, and thank you for all the information, I was able to locate the account with that. I will not need that account number, you provided all the info that I needed. I am looking into those open repair tickets and the service call that is showing on the 31st from my end.

[Edited to remove customer's name]

(edited)

New Contributor

 • 

2 Messages

Thanks John, I appreciate you picking this up.

Two requests, and then a technical note that I think will save your
technician a wasted trip.

First: can the 31st be moved up? We are losing connectivity most days,
including twice this morning. Six more days of that is a lot of lost
work.

Second: could you tell me what the open repair tickets are? I only know
about [Edited: "Personal Information"]. If there are others I would like to see what they say.

The technical note. I have been monitoring this continuously from inside
the network since early August, one ping per second, and the pattern is
consistent. Here is this morning at 9:32 AM, measured at the site:

  Comcast modem              0.00% loss    3.1 ms
  Comcast modem, 1400B ping  0.00% loss    3.5 ms
  Our internal networks      0.00% loss
  Panama City rur301        41.27% loss   42.0 ms
  Google DNS                42.86% loss
  Cloudflare DNS            41.27% loss

The modem answered every probe at 3 ms while your own aggregation router
one hop past it dropped 41% of packets. That is the same shape every
time: modem clean, everything past it failing, our LAN untouched, and a
second Comcast Business site of ours on a different path staying at 0.00%
throughout.

The part I think matters most for scheduling:

I can now reproduce this on demand. Any sustained upload will do it --
this morning it was a routine software push, and the failure appeared
within three minutes. Over nine days of continuous measurement, minutes
with loss are 26 times more frequent while we are uploading than while
we are idle. The uploads in question are capped at 20 Mbps, on a circuit
provisioned for 300 Mbps upstream.

It also fails with nothing running at all -- 75 of the 117 bad minutes
we have recorded happened on an idle line, including a stretch yesterday
afternoon at 4 PM. So load is not required, it just makes it far more
likely.

That combination is why I am asking for a maintenance or plant-side look
at the segment this circuit rides, rather than only a premises visit. A
technician who tests at the demarc while the line happens to be healthy
will find nothing wrong and close the ticket. If he can be here while I
start a transfer, I can very likely make it fail while he watches.

One more data point: I have never measured this circuit above about
83 Mbps upstream against the 300 it is provisioned for, and under a
sustained 20 Mbps the modem's own latency climbs to 40 ms average with
200 ms peaks against a 1.5 ms baseline.

I have exact timestamps for every episode going back three weeks, with
the modem's signal levels recorded alongside them, and I am happy to
hand all of it to whoever comes out.

Thanks again for looking into this.

(edited)