Monitor Internet downtime?

[Elliot B.]

We had a new Internet connection installed a few weeks ago.

 

During the time I've been actively accessing the Web, the connection has dropped at least five times during this time, for around a minute each time.

 

It probably also happens whilst i'm not online (e.g. during the night).

 

I contacted the ISP and they've asked me to swap the Ethernet port that my PC is connected to on the router (load of rubbish as it affects all devices, Wi-Fi devices included) and monitor any drops.

 

Since I'm not actively online 24/7, is there some software that will monitor drops for me?

 

I only have to monitor it for a few days at most and I am happy to keep my PC on 24/7, for the software to monitor things.

 

I've installed DUMeter (along with the web-based service) but couldn't figure out if it logs downtime.

 

Net Uptime Monitor only runs for 30-90 minutes each time before asking you to register it ($10).

 

I went to pfsense's website to download it, and upon choosing x64, I get two options: USB Flash Drive installer, CD Image (ISO) installer. I was hoping for a utility I can simply install in Windows, if at all possible.

 

Any ideas?

Edited December 28, 2017 by Elliot B.

[SnoopZ]

You could setup a broadband quality monitor, this will show you on the graph in red when your connection has gone down and approximately what time as well and how good or bad it may be.

 

https://www.thinkbroadband.com/broadband/monitoring/quality/list 

[Nick H. Supervisor]

ping -t www.google.com > log.txt

 

Set it up and run it in cmd, then use Ctrl+C to stop the recording when you like? It's incredibly basic, but surely that would do the trick?

 

EDIT: There is also this. Although it might do the same thing from the sounds of it.

 

EDIT 2: I guess that only shows the connection from the one computer though, your ISP would still argue that the fault is on your end rather than theirs.

[+Fahim S. MVC]

pfSense is an OS... you could load it into a VM on your PC but seems like a bit of a heavy solution to the problem.

[xendrome]

24 minutes ago, Nick H. said:

ping -t www.google.com > log.txt

 

Set it up and run it in cmd, then use Ctrl+C to stop the recording when you like? It's incredibly basic, but surely that would do the trick?

 

EDIT: There is also this. Although it might do the same thing from the sounds of it.

 

EDIT 2: I guess that only shows the connection from the one computer though, your ISP would still argue that the fault is on your end rather than theirs.

This would work but use fping instead, it can log timestamps to the log file as well - https://fping.org/

[+BudMan MVC]

Why not just do it from the other direction. There are plenty of places that will ping you.. Statuscake been using for a bit...

 

Get lots of great reporting out of it

 

Also alerts - I get an email when my home connection is down.. Plus any of my VPSes, etc. I also run domotz for monitoring all my local devices - which also monitors my internet connect and alerts my on my phone when internet is down, etc.  But its not free - and you need to run it on something. Pi or VM, etc.  https://www.domotz.com/

 

But really slick is I can check my ports traffic on my switches right from my phone   When Im anywhere..

 

While domotz can tell me when stuff down or up - tells me when my son's over picking up my grandkid while I am at work because I see his phone pop up on my network and have it set to alert me on that, etc.

 

[Elliot B.]

It only goes down for 30-60 seconds when it happens. If I got a service to ping me, it would have to do it very often; such as every 5 seconds, all day. It may be a little too much, not sure?

[+BudMan MVC]

Free service pings you every 5 minutes..  But if you pay you can get constant - so prob every second..

 

Pfsense pings its gateway every second.  So you get a graph like this..

 

You could fire up smokeping - free and can give you great graphs of outages and changes in response time..  Is this wired or wireless outages?

 

How exactly do you know it was out - maybe it was just a blip in dns resolution of some time.. etc..  Or a browser hang.

[Circaflex]

11 minutes ago, BudMan said:

Why not just do it from the other direction. There are plenty of places that will ping you.. Statuscake been using for a bit...

 

Get lots of great reporting out of it

 

 

Also alerts - I get an email when my home connection is down.. Plus any of my VPSes, etc. I also run domotz for monitoring all my local devices - which also monitors my internet connect and alerts my on my phone when internet is down, etc.  But its not free - and you need to run it on something. Pi or VM, etc.  https://www.domotz.com/

 

But really slick is I can check my ports traffic on my switches right from my phone   When Im anywhere..

 

 

While domotz can tell me when stuff down or up - tells me when my son's over picking up my grandkid while I am at work because I see his phone pop up on my network and have it set to alert me on that, etc.

 

Those are some nifty services, thanks for sharing Budman. If i had 1/10th of your networking knowledge, I would be happy as a clam.

[Elliot B.]

4 minutes ago, BudMan said:

Free service pings you every 5 minutes..  But if you pay you can get constant - so prob every second..

 

Pfsense pings its gateway every second.  So you get a graph like this..

 

You could fire up smokeping - free and can give you great graphs of outages and changes in response time..  Is this wired or wireless outages?

 

How exactly do you know it was out - maybe it was just a blip in dns resolution of some time.. etc..  Or a browser hang.

I'm using Ethernet but every system in the house (laptops, phones) also no longer get a connection.

 

"Page Cannot Be Displayed"-type messages, with the ISP router "Connection has been lost" display in the browsers.

 

Wait 30-60 seconds and try the page again and it's fine

 

But if you're in VMware Horizon Client, or playing an online game, it'll likely kick you after a 30+ second down spike

[+BudMan MVC]

And when you have such an outage - did you try pinging your routers IP?  Other devices on your network, your switch, etc.

 

You have not shown that is an internet problem.. Could just be your router, or your switch, etc.  Could have a broadcast flood going on, etc.. 

 

Your windows guy.. How about something like spiceworks free network monitoring software?

https://www.spiceworks.com/free-network-monitoring-management-software/

[SnoopZ]

27 minutes ago, Elliot B. said:

I'm using Ethernet but every system in the house (laptops, phones) also no longer get a connection.

 

"Page Cannot Be Displayed"-type messages, with the ISP router "Connection has been lost" display in the browsers.

 

Wait 30-60 seconds and try the page again and it's fine

 

But if you're in VMware Horizon Client, or playing an online game, it'll likely kick you after a 30+ second down spike

Think broadbands monitor pings you every second, you won't notice this at all and this tool is very popular in the UK to look at how reliable your connection is.

https://www.thinkbroadband.com/faq/broadband-quality-monitor

 

Examples of graphs.     https://www.google.co.uk/search?q=think+broadband+graph&source=lnms&tbm=isch&sa=X&ved=0ahUKEwivppu3xa3YAhVNPFAKHRb1AhoQ_AUICigB&biw=1375&bih=862

 

Quote

How does the Broadband Quality Monitoring work?

Once you have registered an IP address for monitoring, we will ping that IP address sending small ICMP echo requests. We send a ping each and every second and average 100 seconds of pings for each point displayed on the graph.

Each ping is very small at only 28 bytes in size, so the average usage per month should be around 73.64 MB (MegaBytes)

 

[+BudMan MVC]

That looks like a winner to me for monitoring your internet... And in the UK too.. every second monitoring - nice!

[Elliot B.]

2 hours ago, SnoopZ said:

Think broadbands monitor pings you every second, you won't notice this at all and this tool is very popular in the UK to look at how reliable your connection is.

https://www.thinkbroadband.com/faq/broadband-quality-monitor

 

Examples of graphs.     https://www.google.co.uk/search?q=think+broadband+graph&source=lnms&tbm=isch&sa=X&ved=0ahUKEwivppu3xa3YAhVNPFAKHRb1AhoQ_AUICigB&biw=1375&bih=862

 

 

This is what I'm seeing:

 

 

My pings are definitely not that high. I play online FPS games a lot and they're below 40 ping at all times.

 

Plus, the graph isn't clearly showing when my connection disappears and for how long.

 

I appreciate your help, though

 

I shall continue my search.

[Circaflex]

If possible, for a few hours or whatever the time frame may be, directly connect your laptop to your modem via an ethernet cable and use as you normally do. If you don't see any drops, chances are it is your router causing the problem. If you see drops, its an ISP issue. Or like Budman mentioned earlier, when you notice the drop, can you ping your router during that time?

[+BudMan MVC]

Your response time there horrible.. The yellow in that graph is the MAX latency in X number of times they pinged you... So 60 pings in a minute.. So there is going to be a min and a max.. They are all not going to be Y ms in response..

 

 

So while it looks bad.. Your min is good, but that high yellow points to widely fluctuation in response.. Could be if you are playing games and using up your pipe then sure your router might not answer back fast as normal.. Or it could be bufferbloat, and just normal queue of your packets if your moving data, etc.  But notice the RED.. on the top - that is not good for sure - that is packet loss..

 

 

Looks like it peaked out at about 5%  Which not good.. But again where you playing games at this time..  But in general your times look good that green is low level and what about 10ms... Blue has a few upswings but other than that one spike where you had packet loss and avg went way up.. 

 

You need to understand what your looking at and you need to zoom in to get more granular..  Did you ping your router local IP from a client when you had this outage?  Did you ping other clients on your network when you had this outage?

 

Also keep in mind that those yellow spikes could be traffic from them to you.. And nothing to do with your actual connection to your isp.. But something going on between where they monitor and you... Need to let that run for longer to see.  I assume that is UK time.. so pretty much that is prime internet time, etc.. Whats the graph look like when internet is more quiet in the uk, and your not doing stuff, etc..

[SnoopZ]

Hopefully you won't have all that yellow if your connection is idle?

 

What i will say is that on the cable Virgin Media Hub3 on 24 downstreams which has the Puma 6 issue it is normal to have the graph rather yellow, but in my experience this doesn't affect my connection but other peoples mileage may vary.

 

[+BudMan MVC]

your avg is not bad there, and your min is stable and don't see any packet loss.. So while that is not the greatest looking graph.. As mentioned already that a packet here or there on a 1 second ping that takes a few extra ms is not all that big of a deal.  Especially when your pinging across multiple hops.. icmp is normally very low priority traffic.. So when a router is passing traffic it doesn't care too much about icmp packets and answering them, etc.

 

If your puma6 has not been updated then yeah that could account for some variation in your ping times sure.. I have a puma6 sb6190 modem - because that is only thing on the approval list from my isp at the current time with my 500/50 service.  Once they move to gig in my area and can go with docsis 3.1 prob switch..  But in my research the firmware has been patched to either greatly reduce the problem if not outright fix it.. if you run the puma6 test on dsl you can get an idea if that is part of the reason for your yellow

 

http://www.dslreports.com/tools/puma6

 

 

You need to run this test to some different servers since you can get varied responses from where the server is your testing too and path to that server, etc.

 

You could also have them run smokeping test to your IP, or their line quality test

http://www.dslreports.com/smokeping

 

But I wouldn't be too worried about the max value unless it was really high like in the 300's or more, etc. Or your avg was all over the place..  yes you would like to see the max be real close to your avg and not spiky.. But the nature of the internet doesn't always allow for that..   What would concern me with the OP graph was that 5% packet loss for a bit of time.. You know simple test would be to just run a 1000 pings to say google (8.8.8.8) what is your min/max/avg..

 

How often do you say this internet goes out?  How about just get a constant ping going with the -t.. Get it running in a window... And then after you have had your outage stop it with contrl C and what does it show for min/max/avg - loss?  Get a few of these going same time different windows to your router IP, other clients on your network... Then after you have one of your outages.. Stop them what do they show for loss of packets?

 

Here I just ran ping to 8.8.8.8 for 200 didn't want to run it for full 1000 to to use as example

 

$ ping 8.8.8.8 -n 200

Pinging 8.8.8.8 with 32 bytes of data:
Reply from 8.8.8.8: bytes=32 time=14ms TTL=54
Reply from 8.8.8.8: bytes=32 time=14ms TTL=54
Reply from 8.8.8.8: bytes=32 time=11ms TTL=54
Reply from 8.8.8.8: bytes=32 time=10ms TTL=54
Reply from 8.8.8.8: bytes=32 time=13ms TTL=54
Reply from 8.8.8.8: bytes=32 time=21ms TTL=54
Reply from 8.8.8.8: bytes=32 time=11ms TTL=54
Reply from 8.8.8.8: bytes=32 time=11ms TTL=54

<snipped>

Ping statistics for 8.8.8.8:
    Packets: Sent = 200, Received = 200, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 9ms, Maximum = 30ms, Average = 12ms

 

So you can see here where max was 30ms, and average was 12..

 

[Elliot B.]

I'll read what you guys have written later.

I just wanted to pop on and give you an updated image:

[+BudMan MVC]

looks better but see packet loss again between 11am and 1130am .. But your max off hours seems to be way closer to your avg, and your avg really close to the min..

 

that is where your max seem to get solid and your avg spikes way up.. Playing a game during this window.. Maxing out your pipe?  Streaming some video, p2p.. download something.. Clearly during this time frame some of the pings were getting lost even not only higher times..

Between 10pm and 8am looks real good though.

 

What are the times your saying you get this outage?  Did you setup the pings to other clients on your network, your router, 8.8.8.8 etc.. while the outage is happening.

 

 

[SnoopZ]

4 hours ago, BudMan said:

your avg is not bad there, and your min is stable and don't see any packet loss.. So while that is not the greatest looking graph.. As mentioned already that a packet here or there on a 1 second ping that takes a few extra ms is not all that big of a deal.  Especially when your pinging across multiple hops.. icmp is normally very low priority traffic.. So when a router is passing traffic it doesn't care too much about icmp packets and answering them, etc.

 

If your puma6 has not been updated then yeah that could account for some variation in your ping times sure.. I have a puma6 sb6190 modem - because that is only thing on the approval list from my isp at the current time with my 500/50 service.  Once they move to gig in my area and can go with docsis 3.1 prob switch..  But in my research the firmware has been patched to either greatly reduce the problem if not outright fix it.. if you run the puma6 test on dsl you can get an idea if that is part of the reason for your yellow

 

http://www.dslreports.com/tools/puma6

 

 

You need to run this test to some different servers since you can get varied responses from where the server is your testing too and path to that server, etc.

 

You could also have them run smokeping test to your IP, or their line quality test

http://www.dslreports.com/smokeping

 

But I wouldn't be too worried about the max value unless it was really high like in the 300's or more, etc. Or your avg was all over the place..  yes you would like to see the max be real close to your avg and not spiky.. But the nature of the internet doesn't always allow for that..   What would concern me with the OP graph was that 5% packet loss for a bit of time.. You know simple test would be to just run a 1000 pings to say google (8.8.8.8) what is your min/max/avg..

 

How often do you say this internet goes out?  How about just get a constant ping going with the -t.. Get it running in a window... And then after you have had your outage stop it with contrl C and what does it show for min/max/avg - loss?  Get a few of these going same time different windows to your router IP, other clients on your network... Then after you have one of your outages.. Stop them what do they show for loss of packets?

 

Here I just ran ping to 8.8.8.8 for 200 didn't want to run it for full 1000 to to use as example

 

$ ping 8.8.8.8 -n 200

Pinging 8.8.8.8 with 32 bytes of data:
Reply from 8.8.8.8: bytes=32 time=14ms TTL=54
Reply from 8.8.8.8: bytes=32 time=14ms TTL=54
Reply from 8.8.8.8: bytes=32 time=11ms TTL=54
Reply from 8.8.8.8: bytes=32 time=10ms TTL=54
Reply from 8.8.8.8: bytes=32 time=13ms TTL=54
Reply from 8.8.8.8: bytes=32 time=21ms TTL=54
Reply from 8.8.8.8: bytes=32 time=11ms TTL=54
Reply from 8.8.8.8: bytes=32 time=11ms TTL=54

<snipped>

Ping statistics for 8.8.8.8:
    Packets: Sent = 200, Received = 200, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
    Minimum = 9ms, Maximum = 30ms, Average = 12ms

 

So you can see here where max was 30ms, and average was 12..

 

I ran this test a while back and i get the same results with the red, i know my Hub3 hasnt been patched, and not really worried about it as i don't have any issues, but here is my graph.

 

Btw i thought i read that you said the Puma6 chipset was impossible to patch and it was just being masked or maybe that was someone else?

 

Anyway this isn't my thread so i will stop posting about my modem, thanks for your help though Budman.

 

 

[+BudMan MVC]

sure wasn't me saying that it was impossible to patch... Here is the thing as you saw my puma test there is some yellow but the spread of response is not all that high..   Here I have pingplotter standard.. Running a test using tcp 80 as the test method and .3 as the interval.. Let it run for a few minutes.  Do you see any huge spikes?

 

 

It its about a 5 ms range for most response.. With min being 12 and max what 28ms tops... That is clearly not what they were saying the issue was before... look over at http://www.badmodems.com/Tools.htm

 

This shows you a sample graph with HUGE spikes in the response time every 30 seconds.. I am currently running 93V for firmware.. Some people say that it only masked the problem by fixing the icmp issue but tcp was still there.  From my testing not seeing it.  And have been very happy with my arris sb6190 and the speed from my new isp..  Will I move to docsis 3.1 -- maybe... But at the time of my switch..  This was the only approved on the list, and I didn't want them coming and saying oh it won't work, or whatever.. Since I did not want to rent from them.. Maybe in a year or so when this has paid for itself in rental fee savings.  I might update to new model - or if sure I start to have issues with the whole puma6 thing.  There are some class actions suites going sure.. Maybe I can get some money from that?  Who knows.

 

But its quite possible they could fix the issue with different firmware - different ISPs have rolled out different versions.  Check yours and see what you have.

 

But not seeing anything like the stuff your seeing with all that red... So you prob on different firmware (older)..

 

What I do recall saying is buying pum6 with the issues know wouldn't be a good thing - and then not shortly after making that statement I went and got one myself   But I was kind of rock and hard place.. Pulled trigger on switch to new isp when got my last isp bill and just got fed up with pay $100 for 75/10 when I could get 500/50 for $59... Especially when they were charging me in the bill for freak tv and all these other fees that took it to $99 when the bill should of been $69... So said F them - and pulled trigger on new isp and install day was only like 2 days out.. So I need a modem and router that could handle the 500 really quick.. Than you amazon prime and same day delivery   Reading their forums and their approved modem list - the 6190 was the only one that was recommended and approved with documentation.  I had read that the 8200 would work which is docssis 3.1 and 3..  But some of the threads  I was reading said it "might" work - depends on area your in for the isp, etc.  So was reading that there was firmware versions out that had corrected some of the problems, etc.  So rolled the dice and sofar been happy with everything. 

 

Grab pingplotter, they have a 2 week trial - but I Just bought it while back its 40$ for standard.. Free version can not do tcp or udp testing.. They have a $7 a month subscription you could do for 1 month if you just want to run some tests for a few days as well.

 

That would prob be create tool for the OP problem... You can have it testing to multiple places at the same time, and can get really fast tests going.. Much faster than 1 ping a second..