No Internet, but can be accessed locally

[karlpox]

Hi,

 

Anyway this is kinda weird but I just experienced 4 blackouts this week. My server was working properly before the blackouts, and now my server doesnt have internet anymore. It can be access through LAN though. Ive tried various changes based on google results but nothing seems to work. Any suggestions? TIA

sobnology-diagnostics-20200831-2049.zip

[Frank1940]

I looked at the Diagnostic file and did not see anything unusual.  What does not have Internet access?   App's update checks, Plugins, Dockers, VM's? 

[karlpox]

10 hours ago, Frank1940 said:

I looked at the Diagnostic file and did not see anything unusual.  What does not have Internet access?   App's update checks, Plugins, Dockers, VM's? 

apparently everything. except for nextcloud. plex,radarr,rutorrent,duckdns and emby are all without internet.

 

cant access the apps, cant check for updates for plugins/docker. I currently dont have vms atm.

 

i also tried doing the new config thing. but didnt work as well.

Edited September 1, 2020 by karlpox

[karlpox]

i changed the mtu to 1500 and it worked. it was working with 9000 for a few months already.

[905jay]

unraid-syslog-20200902-1913.zipI've got a similar issue

 

unRAID and it's containers cannot access the internet. I can't ping out /nslookup /traceroute out.

I was playing with some pfsense firewall settings last night and thought, perhaps I messed something up, somewhere.

So I reverted to the known good working backup, but still no dice.

 

unRAID server has internal connectivity. I can access everything on local IP:PORT

 

piholes have been removed from the network to rule them out, pfsense is giving 8.8.8.8 /1.1.1.1 as the DNS servers

 

I have VMs hosted on unraid that have internet access, and are assigned an IP from pfsense based on MAC address.

I also have some containers hosted on unraid, that are given external SSL access (bitwarden /nextcloud /confluence /homeassistant)

They can all be accessed internally with IP:PORT as well as https://service_name.com 
But I cannot access them from outside the network. 

 

I haven't changed anything on unRAID that I can think of beside enabling an interface on my 4 port NIC, to add to another AP for guest access, but I've reverted all those changes. MTU is set to 1500 per recommendation above 

I have restarted pfSense, unRAID, my AP's just to rule things out, and as mentioned, removed the piholes and am using 8.8.8.8 /1.1.1.1 as DNS for testing to see what the hell is wrong with my setup.

 

Any help would be greatly appreciated.

unraid-syslog-20200902-1913.zip

 

Edited September 2, 2020 by 905jay
added some screenshots

[kana]

I had similar problems that were due to wrong routing using more than one NIC. Can you post a screenshot of the bottom part of the network config with the routing?

What exactly do you see when you ping/traceroute out? Is there a difference between IP adresses and host names?

[905jay]

1 hour ago, kana said:

I had similar problems that were due to wrong routing using more than one NIC. Can you post a screenshot of the bottom part of the network config with the routing?

What exactly do you see when you ping/traceroute out? Is there a difference between IP adresses and host names?

 

The routes highlighted in Yellow, can't be deleted for some reason.

I have docker and vm manager off, and went back into network settings and I can't get rid of it

[ken-ji]

You're seem to have default gateways defined for each subnet / NIC interface. Unless its a router (Unraid is not that capable), don't specify the default gateway for all interfaces. Pick the default/main interface only (usually br0/eth0). Then for VM's on the other interfaces, DHCP can take care of it (or have static settings). Same with dockers, the docker custom network you define for each interface can have the gateway defined there, not on the Unraid level.

 

Just so you don't get confused, I'm running my dockers all on a secondary interface to allow docker and Unraid communications. Its been like this from 6.3 series.

[kana]

56 minutes ago, 905jay said:

 

The routes highlighted in Yellow, can't be deleted for some reason.

 

 

I've had the same problem. Try the command line to delete it. This command may or may not be right, you might have to play around with it:

ip route del default via 10.15.83.1 dev br2

Edited September 2, 2020 by kana
Improved style

[905jay]

2 hours ago, kana said:

I've had the same problem. Try the command line to delete it. This command may or may not be right, you might have to play around with it:

ip route del default via 10.15.83.1 dev br2

@kana thanks very much for your help. I spent the entire day spinning my wheels and couldn't understand where things went wrong.

You've saved my sanity.

 

Should I run anything else to further clean it up?

I have a 4 port NIC

10.15.81.xxx is my Main LAN

10.15.82.xxx is for IOT devices

10.15.83.xxx is for Guest & Family LAN

 

 

 

I am running multiple piholes on unRAID for each of those interfaces (eth0 is Main, eth1 is IOT, eth2 is Guest) so I would like o keep all 3 interfaces connected as I presently have them. I run HomeAssistant which is using br1 (eth1 interface) and I also run Wireguard on my unRAID server and use it frequently.

 

Thanks again for saving my ass, and my sanity. Just seeking your advice so I don't screw anything up

 

 

 

[ken-ji]

@905jay @kana

That's the wrong fix. You want to change your Unraid networking config and not specify default gateways for the other interfaces

Here's my config for my eth1

There's no config at all, as I don't need Unraid to be accessible on that interface, but Dockers can live on that interface and VLAN interfaces without issues.

[kana]

5 hours ago, ken-ji said:

That's the wrong fix. You want to change your Unraid networking config and not specify default gateways for the other interfaces

 

"Not specifying multiple default gateways", what you are actually suggesting, is exactly what my command line accomplishes by deleting superfluous ones, so why do you call it wrong?

Besides, regardless of the issue at hand, the GUI in general seems to have a bug that prevents you from deleting routing entries, for that, my proposed command is a working way indeed and thus not wrong.

[EDIT] Sorry, now I got you, you were referring to 905jay's screenshot with still two gateways in it and not my suggestion. Never mind...

 

@905jay I'm pretty much a beginner myself, so I think @ken-ji 's suggestion to have only one default route in the system and specific routes for the others is perfectly sufficient and does the job.

Edited September 3, 2020 by kana
Added EDIT section

[ken-ji]

Actually the GUI won't let you change routes that are already defined by the system ie result of Network settings.

And the reason its the wrong fix is that it won't stay like that: you restart Unraid; or make networking changes, and then all these extra gateways will all comeback and confuse you. Particularly if you don't reboot until its been months down the line.

[905jay]

thanks for all the great info guys I realy appreciate it.

I'm hesitant to make further changes to the configuration as I host a bitwarden as well as a confluence instance that I rely on for frequent daily use.

I also don't want to not listen to you folks who know what you're talking about.

 

Is it just a matter of me simply leaving the gateway blank for the interfaces eth1 & eth2?

Should I leave it blank for all three interfaces (eth0, eth1& eth2)?

 

I'm not using VLANs or any tagging, just 3 distinct subnets for an intended purpose.

 

Edited September 3, 2020 by 905jay
added some text for clarity

[905jay]

Would assigning the info based on the interface MAC in pfsense make more sense and just set it all to automatic?

That way any gateway issues should be eliminated, right?

 

I don't have a firm understanding of the networking, it's not one of my strengths.

[ken-ji]

Setting it all to automatic would then make your pfsense assign via DHCP to Unraid 3x. For each subnet along with default gateway. So don't. just do static assignment and decide with interface (typically eth0/br0/bond0) is the "main" one and will be used to go out to the internet.

2 hours ago, 905jay said:

Is it just a matter of me simply leaving the gateway blank for the interfaces eth1 & eth2?

Should I leave it blank for all three interfaces (eth0, eth1& eth2)?

Just leave blank for the non-main interfaces

and leaving it blank for all will make your Unraid unable to reach the internet.

[905jay]

You're a beautiful man, Charlie Brown!

I will look into doing this after "production hours" lol
 

Thanks for your help on clarifying this for me @ken-ji 

[905jay]

@ken-ji does this look better to you?

 

 

[ken-ji]

Looks good. I'll assume its working for you.

[905jay]

15 hours ago, ken-ji said:

Looks good. I'll assume its working for you.

yes sir it is working as expected. Thanks for helping me fix this!

[thestip]

On 9/2/2020 at 3:30 PM, ken-ji said:

You're seem to have default gateways defined for each subnet / NIC interface. Unless its a router (Unraid is not that capable), don't specify the default gateway for all interfaces.

 

You have no idea how much time I spent trying to figure out why my server lost 'net access. THANK YOU.

[DuneJeeper]

Hey guys I moved my unraid from an HPE microserver to an r710. Everything booted fine but I can't reach my dockers like nextcloud, etc. from the internet but it's fine on my LAN. I assigned the "new " server the same IP as the old HPE had but that didn't help. Any ideas?