Hi everyone, hoping someone can help me out here. I just got a PiVistation 5 Hawkeye and set it up following the official docs. Right out of the box I’m seeing crazy high CPU temperatures — it hits 85°C within seconds of plugging it in, and the fan is screaming at full speed non-stop. I’m not even running any streaming services or applications, it’s literally just sitting there idling. The thing is borderline unusable because it keeps becoming unstable. I’ve paired it with Home Assistant for a simple periodic snapshot automation and the temps are just as bad. Is this normal? Surely a Pi shouldn’t be cooking itself just taking a few snapshots?
That definitely doesn’t sound right for idle operation. Could you run the following command and share what services are active? That’ll help narrow down what’s eating your CPU:
systemctl list-units --type=service --state=running | grep -E 'camera|stream|libcamera'
Also, what does vcgencmd measure_temp report after a fresh boot with nothing manually launched?
Thanks for the quick reply. So I did some digging on my own before posting and found the camera-streamer service running. I tried disabling it and the temperature dropped — it’s now hovering around 74°C at idle, which is still way warmer than I’d expect but at least it’s not thermal-throttling constantly.
Then I tried enabling libcamera_net instead of camera-streamer to see if that was any better, but nope — straight back to 82°C at idle.
Here are some screenshots of what I’m seeing:
Even with camera-streamer killed, 74°C idle on a Pi that’s just snapping the occasional photo for Home Assistant feels wrong. Is there something else going on?
Thanks for the detailed info and screenshots — that’s really helpful. What you’re experiencing is actually a known issue with the PiVistation 5 Hawkeye out-of-the-box configuration.
Here’s what’s happening: the device ships with camera-streamer video streaming modes fully enabled by default. Even when you’re not actively streaming anything, those services are spun up and chewing through CPU cycles, which is why your temps are through the roof and the fan never settles down. The libcamera_net service behaves similarly — it’s also running in a mode that keeps the CPU pegged.
When you disabled camera-streamer and saw temps drop to ~74°C, that’s the CPU finally getting some breathing room, but there are likely additional streaming-related services still running in the background keeping the baseline higher than it should be.
The short version: the default service configuration needs some manual tweaking to get normal operating temperatures. This isn’t a hardware defect — it’s a software configuration issue with how the streaming services are set up from the factory.
Ah okay, that makes sense. So what’s the fix then? Do I just need to disable all the camera-streamer related services? Which ones specifically, and is there a recommended set of services I should keep running for basic snapshot functionality with Home Assistant? I don’t want to break anything by blindly disabling stuff.
Totally fair question — you don’t want to nuke something that’ll take down your snapshot pipeline. The core issue is that several camera-streamer sub-services are enabled that you simply don’t need for periodic still-image capture. Each one adds to the CPU load.
I’ve put together a step-by-step breakdown of exactly which services are safe to disable while keeping snapshot functionality intact. I’ve sent you the full guide through private message — it includes the specific systemctl commands, which services to mask vs. just stop, and how to verify everything still works with Home Assistant afterwards.
In the meantime, the quick version is: you’ll want to disable the WebRTC and RTSP streaming endpoints while keeping the base camera interface available for libcamera-still or your HA integration. After applying the changes you should see idle temps drop into the 45-55°C range, which is much more reasonable.
Longer term, this is something that needs to be addressed in the default image — having all streaming modes enabled out of the box just doesn’t make sense for most users. But for now, manual service management is the workaround. Let me know how it goes once you’ve applied the changes from the guide!