Strange Networking Issue - Solved

[scott2ware]

I recently transitioned from Fedora 14 to Ubuntu 12.04 when Amahi went stable. Nothing has changed in my setup apart from this and I didn't have this issue previously.

I have DHCP and DNS turned off on my router and use my server to do this which works fine for the most part. When a laptop/tablet/phone etc... connects it gets assigned an address and I can see the lease in the Amahi dashboard. Everything works fine on all devices, I can browse the web, internal network and short names all work fine (/hda etc...).

The problem is that after a few reconnects of the same device or after a couple of days, the device will connect ok and I will be able to browse the internal network and short names work fine, but I am unable to navigate to an external website (google.co.uk etc...).

When this happens other devices which are still connected have internet access no problem at all, only devices using certain IP addresses are affected and the only remedy is to force a different IP address to be assigned on disconnect/reconnect but this has to be done using brute force as by default the same IP gets assigned to a given MAC address.

This issue is really puzzling me but is very frustrating. I will try and give as much information as possible because I appreciate just by reading this it may be hard to understand the problem.

I have disabled the firewall in ubuntu (ufw disable) and all internal web apps/applications/networking/ssh etc... work fine all the time on all devices.

The HDA with it's static IP address has internet access all the time and I have had no issues on this side.

When a client loses internet browsing ability I can still ping an external website but cannot view it in a web browser (DNS issue?). As I previously said, all internal networking still works ("hda" resolves fine).

If it still isn't clear i'll give an example. I connect my phone to my wireless network on Monday and it gets an IP of 192.168.1.101 from my server's DHCP server. I can browse the web and internal network and all seems fine. Over the course of a couple of days I disable and enable wireless on the device several times and each time it gets assigned the same IP and can browse fine. On Tuesday evening I enable wifi on my phone and it gets assigned the same IP address (192.168.1.101) but I cannot browse the internet at all. I can only access internal pages including (***.yourhda.com). The only way I can get the device back on the internet is by getting it assigned a different IP address by the DHCP server. Windows devices still claim they have internet access even though no websites load which once again points to a DNS resolving issue possibly.

This isn't just related to a certain device, most devices on my network have had the problem including laptops, tablets and phones. The only device which hasn't had an issue yet is a tablet which stays connected to wifi constantly. Devices which get turned on and off regularly have all had issues. Any ideas?

EDIT:
So it seems that changing the IP address assigned to a device doesn't always cure the problem. For instance this morning my Ubuntu laptop once again could not load any external websites, only internal but I could ping google.co.uk, amazon.co.uk etc.. So i tried setting a static IP in the amahi dashboard for my laptop, restarting the DHCP server and resetting the wifi on my laptop and still no luck even though the IP address had definitely changed to the static IP I set. So this leads me to believe it is maybe blacklisting certain MAC addresses for a period of time for some reason. On the same laptop I could plug a network cable in (different MAC address) and it gets assigned a different IP by the DHCP server and I instantly have internet access back. Still puzzling me!

[scott2ware]

I am working around the issue currently. When a device cannot browse the web if I change the MAC address of the network card the DHCP server picks it up as a new device and assigns it a new IP address and I can browse the web again. It's a bit annoying having to do this every other day though so it would be good to get to the bottom of it. Why would a device on my network be blocked from forwarding DNS requests but all internal networking and DNS requests work without issue?

[bigfoot65]

Hard to say. Try assigning static IP via dashboard. That's how all my clients are set up.

[ Post made via Android ]

[bigfoot65]

Hard to say. Try assigning static IP via dashboard. That's how all my clients are set up.

Check logs. Did you follow Amahi install docs?

[ Post made via Android ]

[scott2ware]

Thanks for the reply. I did follow the Amahi documentation when installing. I have DNS and DHCP disabled on my router as I did previously with Fedora 14. This has only been an issue since reinstalling Amahi on a fresh Ubuntu 12.04 install. I have tried setting a static IP for a device that has stopped working on my network and occasionally this solves the issue once the static IP is assigned but after a few reconnects the same problem re-appears even with the static IP. I'm at a loss as to what could be blocking a device from forwarding DNS requests after a period of time. As I said before, it is only external websites which can't be accessed (but can be pinged) and ALL internal applications/shares still work. Not all devices are affected at the same time either, sually just one or two whilst the others still work no problem.

Which logs would you recommend checking when the error occurs. I'm not sure which log will show me a device connecting to the network and any errors which go along with it? The only other thing I haven't tried yet is re-enabling the firewall in Ubuntu and setting it up properly rather than leaving all ports open.

[bigfoot65]

All log files are in /var/log. You can also take a look at the hda-install.log in /root to see if it provides any clues.

The problem could be a variety of things. Do you have a firewall enabled on the router by chance?

[scott2ware]

So I have disabled the firewall and intrusion detection on my router. The router intrusion detection had quite a few UDP_Null entries which had been blocked but seem unrelated to this issue. I still had issues after disabling the firewall but it's too early to tell whether disabling intrusion detection has helped.

I have also looked through the logs and tried to get a log of a successful connection vs one which forwarding does not work but it is proving difficult. This is the standard lines I get when a client connects for the first time:

Sep 5 08:15:13 HDA dhcpd: DHCPDISCOVER from 00:24:d7:0d:0f:50 via eth0
Sep 5 08:15:14 HDA dhcpd: DHCPOFFER on 192.168.1.114 to 00:24:d7:0d:0f:50 (scott-Latitude-E6410) via eth0
Sep 5 08:15:14 HDA named[3212]: client 127.0.0.1#51861: view internal: signer "ddnskey" approved
Sep 5 08:15:14 HDA named[3212]: client 127.0.0.1#51861: view internal: updating zone 'home.srw/IN': adding an RR at 'scott-Latitude-E6410.home.srw' A
Sep 5 08:15:14 HDA named[3212]: client 127.0.0.1#51861: view internal: updating zone 'home.srw/IN': adding an RR at 'scott-Latitude-E6410.home.srw' TXT
Sep 5 08:15:14 HDA dhcpd: Added new forward map from scott-Latitude-E6410.home.srw to 192.168.1.114
Sep 5 08:15:14 HDA named[3212]: client 127.0.0.1#39234: view internal: signer "ddnskey" approved
Sep 5 08:15:14 HDA named[3212]: client 127.0.0.1#39234: view internal: updating zone '1.168.192.in-addr.arpa/IN': deleting rrset at '114.1.168.192.in-addr.arpa' PTR
Sep 5 08:15:14 HDA named[3212]: client 127.0.0.1#39234: view internal: updating zone '1.168.192.in-addr.arpa/IN': adding an RR at '114.1.168.192.in-addr.arpa' PTR
Sep 5 08:15:14 HDA dhcpd: added reverse map from 114.1.168.192.in-addr.arpa to scott-Latitude-E6410.home.srw
Sep 5 08:15:14 HDA dhcpd: DHCPREQUEST for 192.168.1.114 (192.168.1.1) from 00:24:d7:0d:0f:50 (scott-Latitude-E6410) via eth0
Sep 5 08:15:14 HDA dhcpd: DHCPACK on 192.168.1.114 to 00:24:d7:0d:0f:50 (scott-Latitude-E6410) via eth0

If the client is known I get:

Sep 5 08:36:34 HDA dhcpd: DHCPREQUEST for 192.168.1.114 from 00:24:d7:0d:0f:50 (scott-Latitude-E6410) via eth0
Sep 5 08:36:34 HDA dhcpd: DHCPACK on 192.168.1.114 to 00:24:d7:0d:0f:50 (scott-Latitude-E6410) via eth0

This looks normal to me. The only errors I can see are the following which can be seen in the log a few times when clients connect but weirdly it doesn't necessarily mean that the client will have forwarding issues:

Sep 5 10:16:38 HDA named[3212]: client 127.0.0.1#44745: view internal: updating zone 'home.srw/IN': update unsuccessful: Cat-E4310.home.srw: 'name not in use' prerequisite not satisfied (YXDOMAIN)
Sep 5 10:16:38 HDA named[3212]: client 127.0.0.1#38222: view internal: updating zone 'home.srw/IN': update unsuccessful: Cat-E4310.home.srw/TXT: 'RRset exists (value dependent)' prerequisite not satisfied (NXRRSET)
Sep 5 10:16:38 HDA dhcpd: Forward map from Cat-E4310.home.srw to 192.168.1.106 FAILED: Has an address record but no DHCID, not mine.
Sep 5 10:16:38 HDA dhcpd: DHCPREQUEST for 192.168.1.106 from 02:13:e9:bd:2c:f6 (Cat-E4310) via eth0
Sep 5 10:16:38 HDA dhcpd: DHCPACK on 192.168.1.106 to 02:13:e9:bd:2c:f6 (Cat-E4310) via eth0

Does this provide any clues at all? I will continue to look for patterns and clues.

[radioz]

Are you certain that you don't have a second DHCP server on your network?
Those do look like strange errors.

[scott2ware]

This may have been a router issue. I definitely had the DHCP server turned off on the router but like most things with an o2 router, this doesn't necessarily mean it has done what you want it to. I used telnet to access the router and disable the dhcp and DNS server flushing both caches and have had no issues since so fingers crossed. Strange that this only seemed to become an issue after installing Amahi for Ubuntu.

[bigfoot65]

Glad to hear you may have solved it. Reason it would be an issue with Amahi is if you have 2 DHCP servers or DNS servers running, eventually they will clash.