IP address mask /32 causes issues with dnsmasq #431

Closed
opened 2025-12-29 01:29:06 +01:00 by adam · 1 comment
Owner

Originally created by @eSoares on GitHub (Feb 15, 2023).

Bug description

I currently run headscale with minor changes from default configuration.
The address space used is 100.64.0.0/10, but each client ends up with an IP address with mask /32 by default.
This mask causes issues with dnsmasq, where it will ignore the interface and not answer correctly dynamic-host queries, since it indicates that address range is a single device instead of a network range of devices.
This issue was pointed where.

Such behavior makes running a DNS server harder and impossible to use a great feature when the server is exposed to multiple networks.

Is there a good reason for that network mask or a way to easily configure the network mask?

Context info

  • Version of headscale used: the one currently in docker latest tag (0.20.0)
  • Version of tailscale client: 1.34.1
  • OS: Linux Raspbian bullseye
  • Kernel version: 5.15.84-v7l+
Originally created by @eSoares on GitHub (Feb 15, 2023). **Bug description** I currently run headscale with minor changes from default configuration. The address space used is `100.64.0.0/10`, but each client ends up with an IP address with mask `/32` by default. This mask causes issues with `dnsmasq`, where it will ignore the interface and not answer correctly `dynamic-host` queries, since it indicates that address range is a single device instead of a network range of devices. This issue was pointed [where](https://github.com/pi-hole/FTL/issues/1531#issuecomment-1430342537). Such behavior makes running a DNS server harder and impossible to use a great feature when the server is exposed to multiple networks. Is there a good reason for that network mask or a way to easily configure the network mask? **Context info** - Version of headscale used: the one currently in docker `latest` tag (0.20.0) - Version of tailscale client: 1.34.1 - OS: Linux Raspbian bullseye - Kernel version: 5.15.84-v7l+
adam added the bug label 2025-12-29 01:29:06 +01:00
adam closed this issue 2025-12-29 01:29:06 +01:00
Author
Owner

@kradalby commented on GitHub (Feb 18, 2023):

I do not think this is something we can do anything about, I believe the Tailscale client will (on most OS/platforms) try to assign the smallest route possible, and I do not think we can do anything about it. I cant seem to find the docs for this tho.

I will close this, but if you find information that we can otherwise do something about it, then feel free to reopen.

@kradalby commented on GitHub (Feb 18, 2023): I do not think this is something we can do anything about, I believe the Tailscale client will (on most OS/platforms) try to assign the smallest route possible, and I do not think we can do anything about it. I cant seem to find the docs for this tho. I will close this, but if you find information that we can otherwise do something about it, then feel free to reopen.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: starred/headscale#431