There are a few reasons why you may want to cache container images, when running workloads in AWS, e.g.:

• Reduce image pull times for large images, from other regions or external registries
• Pulling containers from external registries at a frequency that gets you throttled

When running tailscale up after reboots on my Ubuntu laptop acting as a subnet router, I would get this performance warning:

Warning: UDP GRO forwarding is suboptimally configured on wl0, UDP forwarding throughput capability will increase with a configuration change.

See https://tailscale.com/s/ethtool-config-udp-gro

Following the recommendations didn’t seem to persist the suggested configuration changes after reboots though 🤔

I rebooted my functioning laptop, running Ubuntu 24.04 LTS, Thursday night. When it booted again, it had no detected mouse/trackpad, Wi-Fi, ethernet, Bluetooth or soundcard. Fairly limiting experience. Do not recommend.