Hacker News new | ask | show | jobs
by mercora 1840 days ago
> I would periodically get it working and then it would periodically break.

wait a second... it worked sometimes? without doing any routing trickery or something?

If you did some trickery maybe the devices sometimes chose to speak ipv6 but could not?

2 comments

Oh no, I was trying different bits of routing trickery. I'm pretty sure if I sat down and gave it a solid 3 hours straight of methodical effort, that I'd have figured it out once and for all. Instead, I would have 5-10 minutes total per attempt, try something, see if it worked or didn't, then the next time I tried the Chromecast (possibly weeks later), it wouldn't always be repeatable. I also had the Casts being powered off the TV, so they got hard shutdown and cold-booted pretty often.

In short, I never really cared enough to get it working right as the FireTV was "winning" the convenience battle by enough to make it not matter most of the time and I always had an HDMI cable for the times when I really had to get a screen "sharing" to work.

Did you block Google's DNS IPs? That will break Chromecast in strange ways.
Maybe someday I'll write a rewrite rule to pass google DNS through my PiHole...maybe.

All of this hardcoded DNS server BS in iot devices is a pain.

A warning from someone who did exactly this:

https://news.ycombinator.com/item?id=27214320

TL;DR I tried to force Chromecast (and everything else) to use Pihole as DNS and misbehaving devices (like Chromecast) hammered my pihole into oblivion. I'm talking tens of thousands of requests in a very short amount of time which caused my RPI4 to stop responding to DNS requests (dashboard was still working though). See linked comment thread for details but suffice to say 4 virtual machines with pihole behind two load balancers still saw some downtime. OPNsense gateway is a much better (and safer!) fix IMO :)

You need an mDNS repeater. With that and the appropriate ports opened in the firewall I have cross VLAN casting working perfectly reliably.

OpnSense is the firewall.

that's what i thought too. or somehow having to to relay specific broadcasts or similar ways to make the discovery work. this is why i had the impression it could be related to some routing trickery that might had worked for ipv4 but not for ipv6 and the discovery process could make it appear as they are on the same network segment while link-local ipv6 is available when in fact its not...
Thanks! I’ll try that next time I get an hour or so to mess with it.