Dashboard stops working in less than 1 day

Help
1

Please follow the below template, it will help us to help you!

Expected Behaviour:

The dashboard should show statistics.

Actual Behaviour:

After some hours of bringing up the Pi-Hole the Dashboard stops responding. I have to delete the database in order to make the dashboard work again.

I am using the diginc/pi-hole container. On both R-Pi and on an ubuntu VM. The same result happens on both, whichever I configure as the primary DNS server on my local home network.

I am using both IPv4 and IPv6 on my network and here is an example of the docker-compose I am using for both Pi-Hole instances:

version: '2.1'
services:
pi-hole:
image: diginc/pi-hole-multiarch:debian_armhf
restart: unless-stopped
container_name: pi-hole
cap_add:
- NET_ADMIN
ports:
- "53:53/tcp"
- "53:53/udp"
- "80:80"
volumes:
- /opt/container-config-data/pi-hole/pi-hole:/etc/pihole/
- /opt/container-config-data/pi-hole/dnsmasq.d/:/etc/dnsmasq.d/
environment:
- "ServerIP=192.168.80.110"
- "ServerIPv6=2603:1234:abcd:f0:feed::110"
- "WEBPASSWORD=########"
- "DNS1=75.75.75.75"
- "DNS2=75.75.76.76"
- "INTERFACE=eth0"
- "DNSMASQ_LISTENING=all"
- "TZ=America/Los Angeles"
networks:
pi-hole:
ipv4_address: 192.168.80.110
ipv6_address: 2603:1234:abcd:f0:feed::110
networks:
pi-hole:
driver: macvlan
driver_opts:
parent: eth0

enable_ipv6: true
ipam:
  driver: default
  config:
    - subnet: 2603:1234:abcd:f0::/64
      gateway: 2603:1234:abcd:f0::1
    - subnet: 192.168.80.0/24
      gateway: 192.168.80.1

It is really puzzling why this happens.

Debug Token:

R-Pi Token: ytgwcevy2l
Ubuntu token: 7qk5f64axa

2

Maybe disk runs full ?

df -h

3

Thanks for the sugestion but, the ubuntu VM has a 16G disk which is only 30% full:

sysop@pi-hole:~$ df -h
Filesystem Size Used Avail Use% Mounted on
udev 477M 0 477M 0% /dev
tmpfs 100M 5.8M 94M 6% /run
/dev/mapper/pi--hole--vg-root 15G 3.6G 9.9G 27% /
tmpfs 497M 216K 496M 1% /dev/shm
tmpfs 5.0M 0 5.0M 0% /run/lock
tmpfs 497M 0 497M 0% /sys/fs/cgroup
/dev/sda1 472M 260M 188M 58% /boot
tmpfs 100M 0 100M 0% /run/user/1000

The R-Pi has a 32G SD Card that is only 12% full:

sysop@raspberrypi:~$ df -h
Filesystem Size Used Avail Use% Mounted on
/dev/root 30G 3.3G 25G 12% /
devtmpfs 460M 0 460M 0% /dev
tmpfs 464M 0 464M 0% /dev/shm
tmpfs 464M 47M 417M 11% /run
tmpfs 5.0M 4.0K 5.0M 1% /run/lock
tmpfs 464M 0 464M 0% /sys/fs/cgroup
/dev/mmcblk0p1 63M 22M 41M 35% /boot
tmpfs 93M 0 93M 0% /run/user/1001

These are dedicated devices with no other services on them, other than the docker containers running pi-hole.

Is there something I can look into with the database to see what could be the issue?

Also, I didn't seem to have this issue before enabling IPv6 but, that was a few upgrades back when I enabled v6.

Thanks,

4

What exactly is happening with the web interface? "the Dashboard stops responding" is not very descriptive.

Also check the service status for FTL and dnsmasq:

sudo service pihole-FTL status -l
sudo service dnsmasq status -l
5

I have the same issue. Its like the IP/server is offline... When I ping the IP (192.168.1.101) it comes back as offline... but in my VM VNC I can SSH into it...
https://i.imgur.com/7ITUzFT.jpg
https://i.imgur.com/3urtgbA.png

Thanks

6

Is the device connected to the network via ethernet or wireless? Some wireless devices will go in to powersave mode and drop off the network.

7

ethernet. Cat6...

8

The server is not on the same subnet. Your config sets the IP to 192.168.80.110 but you expect it to be at 192.168.1.101

9

Both processes are running:

root@2731a810a927:/# service pihole-FTL status -l
up (pid 681) 32503 seconds
root@2731a810a927:/# service dnsmasq status -l
up (pid 679) 32506 seconds

One thing I did notice is that due to OSX's IPV6 privacy extensions I am seeing over 1K Clients on the primary PI-Hole.

I am disabling IPV6 on the Pi-Hole's and on my local network to see if the dashboard continues to work for 24 hours.

As for a a more descriptive explanation of the problem - The WebUI would become unresponsive, many times I would get a 503 error when trying to connect and other times the page shows up but with no data. Sometimes I can login to the admin page (without any data showing up) and others when I input the password it would time-out with a 500 error.

In all of the above, deleting/stopping the container and recreating/starting it wouldn't solve the problem.

What would, is deleting the pihole-FTL.db and then restarting the container. Then the dashboard would work again, at least until it stops responding again.

When this happens, requests are still blocked by dnsmasq though.

I'll report back in tomorrow or the next day or so with the results of my test of removing IPV6 from the situation.

10

I'm interested here - what about OSX and IPV6 causes you to see over 1K clients? I run three OSX clients and have not seen this behavior.

11

A few things can influence this:

  • Are your devices hardwired?
  • What OSX version is being used?
  • If Wireless how stable is the Wifi Association?

From my observations, the temporary address assigned by OSX for the link is only considered good for a period of 24 hours per interface or whenever there is a state transition for the interface. All of my OSX devices (iphones, ipads, and macbooks) are wireless and i like in a highly congested wifi community so there are times where I see several temporary addresses on my interface.

For detail, see this article from AnandTech about OSX Sierra: Link.

At a high level apple instituted an ephemeral IPv6 address (starting from Lion) where as the OS will generate a randomly generated IPv6 Address per interface to maintain user privacy, originally created from rfc4941 and likely based on rfc7217. There is alot of history here but long story short IPv6 was created before modern technologies for tracking people existed and this mitigates it.

Keeping true to form of poorly documenting alot of OSX internals, there is very little Official information from Apple on this. The Anandtech article makes some observations about this behavior.

If you just google "OSX sierra IPV6 privacy" you will see many people talking about this feature.

12

I'll do some reading. All my OSX (3) and IOS (5) devices are wireless, all have reserved IP addresses and 24 hour DHCP leases from my router, all are running latest software (OSX 10.13.5 and IOS 11.4.1).

Do you need IPV6 for a particular reason? On a home network, IPV4 works fine, and as far as I know all ISP's are providing IPV4 addresses to your modem. if you switch it off, you'll avoid the overhead and hassle that comes with it.

13

Also, this is for SLAAC. If you are not using SLAAC, then the behavior might be different but I suspect the behavior is the same. I haven't been able to successfully get DHCPv6 working (state-full or stateless) on my firewall so for now its all SLAAC.

I think the temporary address is very good for privacy so I don't want to disable it but have hundreds or thousands of clients in pi-hole for around 20 devices really can be annoying.

14

Have you considered just running IPV4 or will that not support your needs?

15

I work in technology, specifically in the networking field, and having IPv6 is very beneficial for some of my use cases.

At home I will be setting up v4, v6, and dual-stack networks. Ideally Pi-Hole will be on a dual-stack network.

My next step assuming I have no issues with the Pi-Holes running with only v4 my next plan is to move to v6 only to see what the outcome is.

I even looked at the FTL database with sqlite3 and everything looks fine with a superficial inspection so I am leaning towards something with php but will dig into that after I can determine if the issues are directly relate to v6.

16

If you're going to go dual-stack or native IPv6 then the suggested setup is using ULA addressing, while you can use GUA we've found that most ISP's will rotate the allocated address space.

17

Thanks, but I have business internet with a /29 of v4 and a delegated public /60 v6 block that my home firewall cascades to /64's for my home subnets. The addresses do not change.

18

Just to report in, my plan to setup a v6 only subnet is not really feasible without me going through the hoops of configuring NAT64/DNS64 which is more than I want to go through right now.

On other notes, I changed the configuration of my pi-holes (one running on an ubuntu VM 16.04.4 and the other on a rPi-3) to be only IPv4 and they have been stable / fine for 2 weeks.

I will switch back to dual stack for both to see what happens and report back.

1 Like
19

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.