hio77 I am being told to get a new modem. What in the TR's above (or that fact the total disconnections for hours is now sparodic interupptions to packet data ), would indicate the need for a new modem?
The only issue I can see is the shocking first step outside the modem. TR is a one of the most basic of tools and is absolutely vital to ensure we can figure out issues. Ping is one of the most basic uptime checks people run. De-priotising those leads to false warnings for uptime monitors, and makes me wonder what else is depriotised and not working - leading to the issues we face.
OK. so your actual issue is, your being advised to replace the modem.
With these disconnections you are experiencing is there any loss of DSL?
Sounds to me like there is not. which would make a modem being a potential likely cause.
If we got with the assumption it's not a modem fault we then get to the next stages..
Is there spikes of upstream bandwidth? particularly if you have a low upstream connection, this will cause issues that feel like micro studding. delays in packets etc.
There are many valid reasons for deprotization of ICMP, just as there are for no responding at all.
I'm not going to get into that argument any further, as you indicate you administrate networks i'd expect you should know the pros and cons here already. If not, I'd recommend some research :)
How it looks on tools such as pingplotter is a unfortunate side-affect however, even pingplotter supports alternative methods such as unix style (udp) pings that will preform differently.
Most data is downsteam (imap / activesync) to PCs. The VDSL reports great connection speeds when it is running. We aren't on the end of a line by any means. i think our cabinet is 1 - 1.5 km away. We do know that cabinet / area network has given other bigger organizations in the area horrific issues. People closer than us have struggled to make 5Mbps and they are on enterprise level support and have been screwed around for months. They have had outages and and issues for quite a while.
Look at the image below - taken today from the crappy old PC on the long end of the network and via two switches / routers. That's awesome VDSL connectivity. No changes at our end, 1ms jitter.
We set up iperf server at both ends and ran connections - it works great - when it works. 0ms to 20ms between two BP connected systems when it works. When it doesn't there is no response.
I'm old school - started on the internet when all there was for people like me was a single phone line out of Whangarei- and it took 24 hours to get an email to USA and back - as all connections were dial up modem to model / peer to peer via Telix and some of the NZ BBS systems. Tracking failures meant the old tools had to work. Expecting some of the sockets connects like today didn't work.
As it progressed to dial up modems via ISPs then we used ping, telnet etc to track issues. Wrapping monitoring in TCP / IP means that there are lower level checks that don't happen. all socket / UDP / TCP packets do is tell you if it gets there or not. ICMP allows more detailed responses - it is much more informational.
People like google, gateway-nz and other servers respond to icmp traffic - and they are much more vulnerable to smurf etc. Why allow? They know the value of diagnostics.
so far here is what I know for sure:
BP claim we weren't authenticating and had to change all our settings. We didn't but when they reset the connection at their end we got connectivity back - for a time.
We did have DSL.
We have had no information on the results of their checks except when I phoned them back and queried why no news. The reason was it didn't fail for 24 hours. nothing more detailed than that. We have had no detailed info from BP re the issues at the time it was bad and no tests or support to track the cause. As such it is a mystery and all we can do is take wild guesses as we cannot do much in the way of diagnositcs if BP wont engage at a meaningful level.
We cant even find out if the modem tried to authenticate when there were issues. No one grepped logs or bothered to to do more than throw wild guess ideas that will cost a bomb at us.
The only things that have helped are things BP did to their network / at the exchange. I'm still guessing it is their issue.