Duplicate local DNS entries on 2nd import of pre-v6 export

[Nwevyn]

Versions

• Pi-hole: Core version is v6.0.3 (Latest: v6.0.3)
• AdminLTE: Web version is v6.0.1 (Latest: v6.0.1)
• FTL: FTL version is v6.0.2 (Latest: v6.0.2)

Platform

• OS and version: Debian GNU/Linux 12 (bookworm)
• Platform: Proxmox VM

Expected behavior

DNS server to not "fail"

Actual behavior / bug

A clear and concise description of what the bug is.
"DNS SERVER FAILURE" with PIHOLE.LOG (as reported in web interface):
2025-02-22 15:56:38.316 duplicate CNAME at line 123 of /etc/pihole/dnsmasq.conf
2025-02-22 15:56:38.316 FAILED to start up

Reviewing "pihole.toml" both the "hosts" and "cnameRecords" arrays show the newest entries appended to existing entries. A comma is added to the last existing entry so the array is syntactically correct.

Steps to reproduce

Have an existing pre-v6 PiHole server with several local DNS and CNAME records.
On pre-v6 PiHole, use Teleporter page to export existing configuration.

Spin up a new v6 PiHole server. Either bare metal or docker (I've tested both).
On new v6 PiHole use Teleporter page to import the pre-v6 configuration.

On pre-v6 PiHole, change a local DNS entry (either record or CNAME). Not necessary, but it was my use case.
On pre-v6 PiHole, use Teleporter page to export the updated configuration.

On v6 PiHole use Teleporter page to import the updated configuration.

See error reported upon completion of import.
pihole.log page shows above "FAILED to start up" message
Local DNS records (under "settings") show duplicates in both DNS Records and CNAME records
These duplicates are also found in the "pihole.toml" file

Debug Token

• URL: https://tricorder.pi-hole.net/du1nLV3a/

Screenshots

Additional context

Duplicate DNS Records can be deleted on the web page. However, attempting to delete duplicate CNAME records results in " Invalid configuration - dnsmasq: duplicate CNAME at line 122 of /etc/pihole/dnsmasq.conf.temp: " message. These must be deleted from underlying "pihole.toml" file.

A valid backup exported from the v6-PiHole teleporter page can be imported multiple times and will not cause duplicate entries. So, this bug is somewhere in the translation logic.

[DL6ER]

Okay, so the bug path is a v5 Teleporter image imported on v6 does not wipe out the dns.hosts and dns.cnameRecords arrays. While checking the source code, I found that the same issue happens with dhcp.hosts (you probably just didn't have had that). Please try whether

sudo pihole checkout ftl fix/v5_v6_hosts_cnameRecords

fixes this for you.

[Nwevyn]

Ok, I attempted to test this and it didn't work.

Then, I actually read the output of the "checkout" command and noticed:

  [i] Checking for availability of branch fix/v5_v6_hosts_cnameRecords on GitHub
  [✗] Requested branch is not available

  [i] Available branches are:
      - codeql
      - dependabot-npm_and_yarn-development-npm-dependencies-c947a1b127
      - development
      - fix/debugging
      - fix/ede_neterr
      - fix/fork_building_1
      - fix/strange_forwarding
      - help_indent
      - http2
      - master
      - new/GeoIP
      - new/basic_auth
      - new/editorconfig
      - new/libbacktrace2
      - new/libexecinfo
      - new/lsqlite3
      - new/unwinding
      - new/warn_open_resolver
      - special/CI_development
      - test/self-host
      - tweak/adam42
      - tweak/conn_errors
      - tweak/conn_errors_threashold
      - tweak/enabled_total_counts
      - tweak/riscv64-self-hosted
      - tweak/shm_anywhere
      - tweak/two_webthreads
      - update/dnsmasq

I don't see anything in the available list that looks like what you requested (assuming a possible typo). So, not sure what happened.

[Nwevyn]

Now that I'm back from my morning appointment, I did a little more research here. I noticed that this fix had been merged into development and the patch was deleted. So, I took a chance (on my test VM) and did a checkout of the "development" branch and retested.

And this did fix my issue.

Thanks for all your effort.