James Tucker
James Tucker
Tailscale is slow: `wg: Failed to write packets to TUN device: write /dev/net/tun: invalid argument`
there does not appear to be anything relevant in https://gitlab.archlinux.org/archlinux/packaging/packages/linux-lts recent history, or local patches
Tailscale is slow: `wg: Failed to write packets to TUN device: write /dev/net/tun: invalid argument`
After some digging, so far it looks likely that the culprit is https://github.com/gregkh/linux/commit/e269d79c7d35aa3808b1f3c1737d63dab504ddc8 and the fix is https://github.com/gregkh/linux/commit/89add40066f9ed9abe5f7f886fe5789ff7e0c50e Next up, verify this.
Tailscale is slow: `wg: Failed to write packets to TUN device: write /dev/net/tun: invalid argument`
We have a potential mitigation for this issue in our wireguard-go which appears to avoid the issue in my lab environment, however as the issue was not triggering so consistently...
Tailscale is slow: `wg: Failed to write packets to TUN device: write /dev/net/tun: invalid argument`
> Mhh, I haven't seen anything like that in dmesg. > > Since I just downgraded my LTS kernel and put it in hold, I tried the whole thing with...
Tailscale is slow: `wg: Failed to write packets to TUN device: write /dev/net/tun: invalid argument`
Ok, there's still some additional behavior that we need to reproduce and understand here, I am not able to completely reproduce the case that @marek22k has yet. I was able...
Tailscale is slow: `wg: Failed to write packets to TUN device: write /dev/net/tun: invalid argument`
> You folks do any testing for packet loss after this? I see a ton after applying to 6.6.44 for an openwrt appliance. Heads up. The message is suppressed but...
If you arrive here because you were searching about some symptoms, but are not in a situation that matches the ticket description, please reach out to support: https://tailscale.com/contact/support#support-form or file...
the udp bind is separate from the control plane connection, which is what was failing in the health check. ``` default via 31.10.245.25 dev eth0 proto static metric 10 ```...
> [@ociaw](https://github.com/ociaw) The idea about the SIGPIPE handler is to get correct error codes for failing pipe situations. We especially had in mind a `command | rustic backup -` where...
Yes, it was eMMc, so likely a dup!