Every 10-15 minutes, my Internet connection fails in the following manner:

Cannot load websites
Cannot connect to Dropbox
Cannot connect to IRC
Skype still works
Slack still works
Still able to connect to my router and modem

After much searching, I believe this is a DNS issue. I have tried using both my ISP's DNS servers and Google's DNS servers, but the issue remains.

When I disconnect from my Wi-Fi network and reconnect, the issue goes away and everything works for another 10-15 minutes.

Here is some output from a few ping tests when the issue appears:

# ping 8.8.8.8 (Google's DNS server, becomes unreachable)

64 bytes from 8.8.8.8: icmp_seq=11589 ttl=41 time=61.719 ms
64 bytes from 8.8.8.8: icmp_seq=11590 ttl=41 time=61.869 ms
64 bytes from 8.8.8.8: icmp_seq=11591 ttl=41 time=60.212 ms
64 bytes from 8.8.8.8: icmp_seq=11592 ttl=41 time=60.332 ms
64 bytes from 8.8.8.8: icmp_seq=11593 ttl=41 time=65.169 ms
64 bytes from 8.8.8.8: icmp_seq=11594 ttl=41 time=61.890 ms
64 bytes from 8.8.8.8: icmp_seq=11595 ttl=41 time=59.746 ms
64 bytes from 8.8.8.8: icmp_seq=11596 ttl=41 time=60.221 ms
Request timeout for icmp_seq 11602
Request timeout for icmp_seq 11603
Request timeout for icmp_seq 11604
Request timeout for icmp_seq 11605
Request timeout for icmp_seq 11606
Request timeout for icmp_seq 11607
Request timeout for icmp_seq 11608
Request timeout for icmp_seq 11609

# ping 203.144.206.49 (ISP's DNS server, automatically configured, becomes unreachable)

64 bytes from 203.144.206.49: icmp_seq=1418 ttl=249 time=27.160 ms
64 bytes from 203.144.206.49: icmp_seq=1419 ttl=249 time=23.846 ms
64 bytes from 203.144.206.49: icmp_seq=1420 ttl=249 time=25.674 ms
64 bytes from 203.144.206.49: icmp_seq=1421 ttl=249 time=25.712 ms
64 bytes from 203.144.206.49: icmp_seq=1422 ttl=249 time=25.169 ms
64 bytes from 203.144.206.49: icmp_seq=1423 ttl=249 time=24.310 ms
64 bytes from 203.144.206.49: icmp_seq=1424 ttl=249 time=26.983 ms
64 bytes from 203.144.206.49: icmp_seq=1425 ttl=249 time=26.477 ms
Request timeout for icmp_seq 1428
Request timeout for icmp_seq 1429
Request timeout for icmp_seq 1430
Request timeout for icmp_seq 1431
Request timeout for icmp_seq 1432
Request timeout for icmp_seq 1433
Request timeout for icmp_seq 1434
Request timeout for icmp_seq 1435

# ping 192.168.1.1 (modem, remains reachable)

64 bytes from 192.168.1.1: icmp_seq=1760 ttl=64 time=1.571 ms
64 bytes from 192.168.1.1: icmp_seq=1761 ttl=64 time=1.414 ms
64 bytes from 192.168.1.1: icmp_seq=1762 ttl=64 time=1.421 ms
64 bytes from 192.168.1.1: icmp_seq=1763 ttl=64 time=1.439 ms
64 bytes from 192.168.1.1: icmp_seq=1764 ttl=64 time=1.600 ms
64 bytes from 192.168.1.1: icmp_seq=1765 ttl=64 time=2.117 ms
64 bytes from 192.168.1.1: icmp_seq=1766 ttl=64 time=1.354 ms
64 bytes from 192.168.1.1: icmp_seq=1767 ttl=64 time=1.395 ms
64 bytes from 192.168.1.1: icmp_seq=1768 ttl=64 time=1.492 ms
64 bytes from 192.168.1.1: icmp_seq=1769 ttl=64 time=1.326 ms
64 bytes from 192.168.1.1: icmp_seq=1770 ttl=64 time=1.641 ms
64 bytes from 192.168.1.1: icmp_seq=1771 ttl=64 time=1.428 ms
64 bytes from 192.168.1.1: icmp_seq=1772 ttl=64 time=1.459 ms
64 bytes from 192.168.1.1: icmp_seq=1773 ttl=64 time=1.517 ms
64 bytes from 192.168.1.1: icmp_seq=1774 ttl=64 time=1.429 ms
64 bytes from 192.168.1.1: icmp_seq=1775 ttl=64 time=2.007 ms

And here is a traceroute both when the connection works and when it does not:

# traceroute 8.8.8.8 (connection is working)

traceroute to 8.8.8.8 (8.8.8.8), 64 hops max, 52 byte packets
 1  192.168.1.1 (192.168.1.1)  1.314 ms  3.256 ms  1.089 ms
 2  cm-134-196-10-1.revip18.asianet.co.th (134.196.10.1)  9.022 ms  9.922 ms  9.988 ms
 3  10.92.249.49 (10.92.249.49)  23.733 ms  16.544 ms  17.930 ms
 4  203-144-128-34.static.asianet.co.th (203.144.128.34)  23.399 ms  22.948 ms  23.950 ms
 5  203-144-128-33.static.asianet.co.th (203.144.128.33)  23.067 ms
    203-144-128-29.static.asianet.co.th (203.144.128.29)  25.810 ms
    203-144-128-33.static.asianet.co.th (203.144.128.33)  23.437 ms
 6  61-91-213-177.static.asianet.co.th (61.91.213.177)  25.623 ms  23.378 ms  24.319 ms
 7  61-91-213-35.static.asianet.co.th (61.91.213.35)  26.058 ms  26.429 ms  31.222 ms
 8  61-91-213-81.static.asianet.co.th (61.91.213.81)  25.335 ms  25.126 ms  23.935 ms
 9  tig-net25-61.trueintergateway.com (122.144.25.61)  24.232 ms
    tig-net25-105.trueintergateway.com (122.144.25.105)  27.276 ms
    tig-net25-209.trueintergateway.com (122.144.25.209)  28.039 ms
10  72.14.195.115 (72.14.195.115)  49.303 ms  49.605 ms  50.321 ms
11  209.85.242.240 (209.85.242.240)  49.322 ms  50.768 ms  49.716 ms
12  209.85.242.242 (209.85.242.242)  58.872 ms  60.480 ms
    209.85.242.232 (209.85.242.232)  67.498 ms
13  209.85.246.23 (209.85.246.23)  62.638 ms
    209.85.248.25 (209.85.248.25)  60.055 ms  60.914 ms
14  * * *
15  google-public-dns-a.google.com (8.8.8.8)  61.586 ms  60.368 ms  61.882 ms

# traceroute 8.8.8.8 (connection is NOT working)

traceroute to 8.8.8.8 (8.8.8.8), 64 hops max, 52 byte packets
 1  * * *
 2  * * *
 3  * * *
 4  * * *
 5  * * *
 (it goes on like this until the connection kicks in again)

Any ideas how to solve this issue?

Graham Swan

Posted 2015-05-18T04:46:57.880

Reputation: 280

Answers

Solution (May 24, 2015)

The unstable connection turned out to be a problem with Mac OS X Yosemite and is apparently a common thing. There are numerous potential solutions to this problem posted all over the internet, but the one that worked for me was found in this Apple discussion thread.

Solution

Move your /Library/Preferences/SystemConfiguration folder to your Desktop (so you have a backup) and reboot. OS X will regenerate default network settings upon reboot.

sudo mv /Library/Preferences/SystemConfiguration ~/Desktop
sudo shutdown -r now

Edit (Nov 8, 2016)

I've moved since posting this question and the problem followed me to my new home (different country, different ISP). I noticed that I could use my laptop on other people's Wi-Fi with no problems, but as soon as I returned home, the problem surfaced again.

The unstable connection turned out to be a problem with the cheap routers some ISPs provide.

My former ISP provided a poorly-reviewed Technicolor unit and my current ISP provided an ancient Cisco unit. As soon as I purchased a decent router, the problem disappeared immediately and has not returned since switching to the new router 2 months ago.

Solution

Buy a decent router and use that for your Wi-Fi.

For reference, the router I purchased was the ASUS RT-AC68U: https://www.asus.com/us/Networking/RTAC68U/

Graham Swan

Posted 2015-05-18T04:46:57.880

Reputation: 280

Thanks for the update, but your previous answer was useful though (not everyone can easily change his router). Please can you bring your previous answer back and append the current answer as an "EDIT:" – ncuillery – 2016-12-29T14:38:48.273

1@ncuillery That's true that not everyone is able to replace their router. I've re-added the solution from May 24, 2015 and changed the latest answer to an edit. Good call! – Graham Swan – 2016-12-30T08:42:59.627

At a guess, I'd say it's a router problem. Make sure you have the latest firmware installed or try another router that is known to work.

JPain

Posted 2015-05-18T04:46:57.880

Reputation: 91

My ISP came by yesterday and replaced the cable modem, but that problem remains. I'm using an Apple Airport Express as my router and it has never exhibited problems on any other connection. Is there any specific setting on the cable modem that may cause this? – Graham Swan – 2015-05-19T14:35:32.897

@thinkswan ok, my bad. If it was a DNS issue, the ping to the ip wouldn't be interrupted. Even if it was a DNS issue, DNS requests are cached many times by many things. – JPain – 2015-05-19T15:15:29.400

Another thing I've noticed is that if the connection dies on one device, other devices on the network can still load webpages, so perhaps it's an issue with the actual device settings. Bah, the investigation continues. – Graham Swan – 2015-05-20T05:02:05.040

Here is a small script I use when I encounter this problem:

#!/bin/sh

while [ true ]
do

    ping -W 500 -c 1 192.168.1.1

    if [ $? -eq 2 ]
    then
        arp-scan -l -I en0
    else
        sleep 1
    fi
done

I hope this can help some of you.

Linvi

Posted 2015-05-18T04:46:57.880

Reputation: 101

I had problems with disconnects, slow internet and modem problems so here are the things I did:

Prior to Nov 2014 I had a SB6121 modem and comcast Blast 50/10 and don't recall any problems with disconnects or speed.
In Nov 2014 (I think) I upgraded to extrem 105 and started having disconnect problems randomly (bad modem??)
Jan 2015 upgraded modem to SB6141. Still having disconnect problems randomly (worst than SB6121) lots of t4 timeouts on upload channel 3 plus other errors
April or May I had comcast tech come and check things out. The tech said he could not see any problems on their end but was unable get a comcast modem to work any better so he restalled the SB6141 and left. (Cost me $70) Still had random disconnects. Maybe bad modem???

05/20/2015 installed a Zoom 5341J modem. Checked status page and found only 4 of the 8 downstream channels were bonded but internet was Ok yet very high uncorrectable code words.

 Downstream Bonded Channels
1   QAM256  621000000 Hz    -0.8 dBmV   39.8 dB 615 1643
2   QAM256  615000000 Hz    -1.3 dBmV   39.4 dB 810 1634
3   QAM256  627000000 Hz    -0.1 dBmV   39.9 dB 522 1520
4   QAM256  633000000 Hz    -0.6 dBmV   39.9 dB 520 1916
5   unknown         0 Hz    -0.0 dBmV    0.0 dB   0    0
6   unknown         0 Hz    -0.0 dBmV    0.0 dB   0    0
7   unknown         0 Hz     0.0 dBmV    0.0 dB   0    0
8   unknown         0 Hz     0.0 dBmV    0.0 dB   0    0

Upstream Bonded Channels

1   ATDMA   5120 Ksym/sec   29500000 Hz 46.8 dBmV
2   ATDMA   5120 Ksym/sec   36400000 Hz 37.5 dBmV
3   ATDMA   5120 Ksym/sec   22600000 Hz 36.5 dBmV
4   Unknown  0 Ksym/sec        0 Hz  0.0 dBmV
Total Correctables  Total Uncorrectables
           2467           6713

Current System Time: Wed May 20 08:15:48 201

Did a comcast chat session to find out why only 4 channels bonded instead of 8 and was told the modem was perhaps set as 5341 not a 5341J and needed to be reactivated so I needed to call comcast. I did, and finally after 30 min or more on the phone the tech said I should see changes within 24 hr. An hour later I checked the status page and found all 8 channels bonded. No internet problems.
Replaced all cables from drop outside to the modem with RG6 cable. Found the old cable had 2 splice connectors in the line. Just making sure the cables were not causing any problems.

05/21/15 AM, Strange to me but I noticed the downstream power levels were very high +12db to +16db yet prior to cable replacement the levels were as above. It seems the change was probably due to cable replacement so I added a 12db attenuator to the drop and this brought the power levels down to:

Downstream Bonded Channels
1   QAM256  591000000 Hz    -2.3 dBmV   39.4 dB 38  195
2   QAM256  597000000 Hz    -2.1 dBmV   39.4 dB 0   0
3   QAM256  603000000 Hz    -1.1 dBmV   39.9 dB 0   0
4   QAM256  609000000 Hz    -0.1 dBmV   39.9 dB 0   0
5   QAM256  615000000 Hz    -0.1 dBmV   39.8 dB 0   0
6   QAM256  621000000 Hz    -0.1 dBmV   39.9 dB 0   0
7   QAM256  627000000 Hz     0.4 dBmV   39.9 dB 0   0
8   QAM256  633000000 Hz     0.2 dBmV   39.5 dB 0   0

The upstream power levels seems a bit high to me (probably due to the attenuator) but within specs

Upstream Bonded Channels

1   ATDMA   5120 Ksym/sec   22600000 Hz 44.5 dBmV
2   ATDMA   5120 Ksym/sec   29500000 Hz 46.0 dBmV
3   ATDMA   5120 Ksym/sec   36400000 Hz 46.0 dBmV
4   Unknown  0 Ksym/sec        0 Hz  0.0 dBmV

05/21/15 afternoon, so far no internet problems other than uncorrectable codewords (195) Not sure if this will be a problem.

New status page results:

Downstream Bonded Channels
1   QAM256  591000000 Hz    -2.3 dBmV   39.4 dB 38  195
2   QAM256  597000000 Hz    -2.0 dBmV   39.5 dB 0   0
3   QAM256  603000000 Hz    -1.1 dBmV   39.8 dB 0   0
4   QAM256  609000000 Hz     0.0 dBmV   40.2 dB 0   0
5   QAM256  615000000 Hz    -0.1 dBmV   39.9 dB 0   0
6   QAM256  621000000 Hz    -0.2 dBmV   39.9 dB 0   0
7   QAM256  627000000 Hz     0.3 dBmV   39.9 dB 0   0
8   QAM256  633000000 Hz     0.2 dBmV   39.9 dB 0   0


Upstream Bonded Channels

1   ATDMA   5120 Ksym/sec   22600000 Hz 44.5 dBmV
2   ATDMA   5120 Ksym/sec   29500000 Hz 46.0 dBmV
3   ATDMA   5120 Ksym/sec   36400000 Hz 46.0 dBmV
4   Unknown  0 Ksym/sec        0 Hz  0.0 dBmV

Speedtest result 111 down 23.41 up using wireless connection from R8000 router 40ft away. Happy so far but at this point I'm not very confident that it will stay stable. If not I would suspect a problem with the line to the pole or line to comcast head end. Just guessing but time will tell.

05/22/2015 The event log is empty (great) and speedtest result 118.4 down 23.4 up

The connection status as of this morning, uncorrectable code words are higher but my son played world of tanks for more than 5 hours while my grandson played Minecraft and lots of youtube clips for 6 or more hours. At the same time myself and my wife were both on the internet and had a netfilx movie playing at the same time. No one complained of any problems, so far so good.

Downstream Bonded Channels
1   QAM256  591000000 Hz    -2.2 dBmV   39.6 dB 539 2770
2   QAM256  597000000 Hz    -2.0 dBmV   39.8 dB 202 957
3   QAM256  603000000 Hz    -1.1 dBmV   39.9 dB 0   0
4   QAM256  609000000 Hz    -0.1 dBmV   40.3 dB 0   0
5   QAM256  615000000 Hz    -0.1 dBmV   39.8 dB 0   0
6   QAM256  621000000 Hz    -0.1 dBmV   39.9 dB 0   0
7   QAM256  627000000 Hz     0.4 dBmV   40.0 dB 0   0
8   QAM256  633000000 Hz     0.2 dBmV   39.9 dB 0   0


Upstream Bonded Channels

1   ATDMA   5120 Ksym/sec   22600000 Hz 44.5 dBmV
2   ATDMA   5120 Ksym/sec   29500000 Hz 46.0 dBmV
3   ATDMA   5120 Ksym/sec   36400000 Hz 46.0 dBmV
4   Unknown    0 Ksym/sec          0 Hz  0.0 dBmV

WilliamF

Posted 2015-05-18T04:46:57.880

Reputation: 1

Thanks for your help with this. After much searching and reading, I was able to solve the problem and have posted the solution below. – Graham Swan – 2015-05-24T09:11:26.630

Wi-Fi stops working every 10 minutes on Mac OS X

Answers

Solution (May 24, 2015)

Edit (Nov 8, 2016)