Health Check Failing (a tale of two modems)


Can put in a switch in between the devices to isolate the physical connection issue ?

1 Like

I removed a splitter from the cable modem which improved signal levels (downstream is now about -1.5dBmV and upstream about 45). I’ll report back if this stops modem 1 from dropping out.

What does your ISP say when they remote their remote diagnostics, I wonder?

Removing the splitter (which improved the Modem’s signal levels) did not fix the problem, as I had another disconnect today.

Next, I will try swapping out the ethernet cable.

If it made it three days with only one disconnect, I would say you are on the right path by looking at the cabling. If it were me, I would ask the ISP to come out and do signal tests on the line that you are using and get it perfect. They have all the fancy gadgetry and knick-knacks to get it right in the sweet spot.

1 Like

jmjones: Agreed … which is why I had suggested earlier for soylent to contact the ISP and ask them to check the levels. DOCSIS provides a means to do this remotely. When I reread this thread the thing that still jumps out at me is a “level issue.” Adding and deleting splitters/distribution amps/whatever is interesting but not non-conclusive as to any issue, really. A remote diag will answer the question. (More signal is not necessarily better.)
I’ve seen this time and time again: The levels, forward or reverse path, are at the margins of acceptability and reliability suffers.
If I had some extra time on my hands, the next thing I’d do is conduct a search for any of the older type crimp-on “F” connectors and any connectors which can be twisted on the cable. All connectors should be the newer compression type and should be resistant to movement on the cable.

1 Like

Had another disconnect about 12 hours after the last one, so the improved signal levels seem to have made no change.

The reason I have not contacted the ISP is that signal levels are just about perfect now (with consistent SNR and Power levels across all downstream/upstream bands, with very few Correcteds and zero Uncorrectables). There’s zero correlation between events in the modem’s log and these disconnects reported by the Peplink. Also, my other modem (CM600) does not show this problem at all, whereas both the current DG1670 and a prior DG1670 show the behavior.

I’ll try a new ethernet cable next, just to be systematic about eliminating variables.

Ok, just had a worse failure on WAN1. This time HealthCheck failed and I tried to go to to get the modem stats, and couldn’t reach it. Ping to also not working. I checked on the peplink support.cgi page and it indicated all ethernet ports were connected at 1000mbps.

I went to physically look at the modem and peplink, and all lights were working normally (both DOCSIS connection as well as the ethernet ports on both devices).

I used the chance to swap in a new Cat 5 ethernet cable and waited a bit, and nothing changed.

Then, I rebooted the modem, and after a few minutes everything came back online.

So this is very interesting - some sort of failure happens in which packets from the Peplink to/from the modem are not being delivered at all.

Seems like a bug in the Arris DG1670 hardware or firmware to me, right?

Also right after the modem rebooted the ping results to it were very poor: Pings would work for 10-20 seconds then fail for 10 seconds.

70 packets transmitted, 145 packets received, 14.7% packet loss
round-trip min/avg/max/stddev = 2.009/7.749/99.222/9.275 ms

Finally it settled down, but however, the ping times are still terrible and there is still packet loss:

--- ping statistics ---
100 packets transmitted, 99 packets received, 1.0% packet loss
round-trip min/avg/max/stddev = 1.918/7.231/34.746/4.474 ms

1% packet loss? 34.7msec maximum ping time? That’s terrible!

This is over wired Gigabit Ethernet, from my laptop, to the peplink, to the modem, with basically zero traffic on the LAN or WAN. Ping times should be sub 1msec and there really should not be any dropped packets.

Here is pinging the peplink itself:

100 packets transmitted, 100 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 0.385/0.516/0.634/0.063 ms

That’s about what I’d expect from Gigabit ethernet.

The peplink had 62 days of Uptime, so I rebooted it and repeated the ping to the modem:

--- ping statistics ---
100 packets transmitted, 100 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.819/6.837/26.615/3.801 ms

No real change there.

So, the problem seems to be happening in the ethernet connection between the Peplink and modem Wan 1.

Sounds like the modem is crapping out on you bud. You can try to change to a different WAN port on the Peplink if you think it could be a bad interface on the router.

I’ve been seeing a similar problem on my Balance One with two different ISPs. One thing you can do is call the ISP’s tech support and have them monitor the services then call them again once your Peplink has logged more errors to see if the ISP sees them as well. After doing this myself, (and having one of the ISPs swap out their modem) it’s looking like the problem is in my Balance One. But I also recently changed the LAN port on the ISP routers that the Balance One was connected to (to rule out the physical port as a problem) and my situation has improved, though it may be coincidental. I have to let it run for a couple days to be sure.

Anyway, hope some of these ideas help you narrow things down.

A comparison of my two modems:

Arris DG1670 on WAN1

— ping statistics —
100 packets transmitted, 100 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.819/6.837/26.615/3.801 ms

Netgear CM600 via WAN2:

--- ping statistics ---
100 packets transmitted, 100 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 1.065/1.216/1.630/0.092 ms

Just for prosperity’s sake. What happens when you plug a laptop in directly to the modem and run the same ping test? If the modem is slow responding to pings without the Peplink involved, perhaps the interface on the modem has gone bad.

1 Like

Word on the street is that the DG1670 may suffer from the Puma-6 chipset problem as does other modems such as the SB6190: Arris Tells us It's Working With Intel on SB6190, Puma6 Problems | DSLReports, ISP Information and that there may be firmware fixes coming: [ALL] SB6190 and CM700 latency fix firmware ready to test. - Cox | DSLReports Forums

but that I may be better off asking my ISP for a different modem in the short term.

1 Like