Chain Manager screen lock up on Vangelis (Resolved)

spurkopf · May 5, 2026, 10:08am

That’s absolutely fine, just an apology in case my answer is nonsensical.

Log in via SSH and delete the folder?

You mean that could help?

wyleu · May 5, 2026, 10:18am

No, I don’t think it’s an SSD issue, per se.

I’ll burn a clean SSD from something pretty recent and give the backup/restore mechanism some use on a completely new build. Vangelis is still R&D and the occasional rebuild is part of the journey for me, cos I muck around a lot. But I do like to try to breath a bit of life into a dying system just to see if something can be learnt.

Since this is on the cajon and it goes out to do drumming circles occasionally, there is the basic reverb on internal mic rig to preserve.
I tend to run cutting edge in live to keep the developer’s honest and see what can be achieved. But once in a while the machine needs to be re built from scratch, based on something I’ve done. My rack machine is up to rack7, but still maintains a large library of material, that I use in various scenarios.
It works for me, but I wouldn’t recommend it for others.

hannesmenzel · May 5, 2026, 12:30pm

I wonder if running this script would help:

github.com/zynthian/zynthian-sys

scripts%2Frecipes%2Finstall_TheUsualSuspects_prebuilt.sh

vangelis

#!/bin/bash

BASE_URL_DOWNLOAD="https://os.zynthian.org/plugins/aarch64"

cd /usr/local/lib/lv2 || exit

if [ -d "Osirus.lv2" ]; then
	rm -rf Osirus.lv2
fi
if [ -d "OsTIrus.lv2" ]; then
	rm -rf OsTIrus.lv2
fi
if [ -d "Vavra.lv2" ]; then
	rm -rf Vavra.lv2
fi
if [ -d "Xenia.lv2" ]; then
	rm -rf Xenia.lv2
fi
if [ -d "Nodal2X.lv2" ]; then
	rm -rf Nodal2X.lv2

This file has been truncated. show original

The path seems odd as well…

Aethermind · May 5, 2026, 2:10pm

I don’t know if the issue is somehow related, but - as reported before in the general Vangelis thread -, I also got irregular behaviour from TUS plugins after one of the latest updates. Namely, Osirus ROM banks and presets aren’t shown anymore on Zynthian (even after regenerating them from developers repo), and some Osirus presets seem to suffer from random and non-reproducible instability of certain parameter settings.

Best regards

wyleu · May 5, 2026, 4:17pm

Well a new SSD and a rebuild to ORam seems to have brought the cajon back to life….

jofemodo · May 6, 2026, 9:11am

This looks like a “partially downloaded file” in the wrong place, probably due to a “broken” install recipe.

You should delete the file, mate:

rm -f "/usr/local/lib/lv2/TheUsualSuspects.tar.xz.1"

Regards,

wyleu · May 6, 2026, 6:02pm

Rebuilt a clean cajon ( cajon5 as Oram) (cajon6 updated to Vangelis)

Oram works ok

On cajon6, Vangelis crashes without even a debug whimper on selecting Chain Manager from Main Menu.
Everything else on the Menu does sensible stuff like what it’s meant to.

Rack7 works correctly not a metre away is running precisely the same hashes as the cajon machine.

In more terrifying news I’m back onto the zynthian wiki . . . .

smespresati · May 6, 2026, 8:48pm

On cajon6, Vangelis crashes without even a debug whimper on selecting Chain Manager from Main Menu.

Hi, @wyleu

I think this issue is similar to the one reported here. Perhaps you can try

At this point Vangelis worked for me.

Best regards.

Aethermind · May 7, 2026, 5:15am

Hi @wyleu,

As @smespresati also points out, you might have noticed that a specific thread was focused on troubleshooting the non-functional behaviour of certain new display and touch drivers of Vangelis, when Chain Manager is triggered on some custom screen hardware.

@riban found that freezing Vangelis to a previous development stage with a targeted uninstall command, prior to the introduction of the toggable keypad overlay (and some other kernel changes I guess), was a temporarily viable solution.

Best regards

wyleu · May 7, 2026, 9:52am

Upon crash one of the threads tops out….

Aethermind · May 7, 2026, 4:46pm

Hmmm… Interesting:

the single but paramount python process

zynthian_main.py

seems to spike at a whopping 101% usage of the CPU Core 0, while occupying just a mere 7% of the available memory, and has kept on doing so for more than ten minutes, thus effectively being frozen/locked.

riban · May 7, 2026, 4:59pm

This is indicative of a tight loop in the code. I am still trying to reproduce this.

riban · May 8, 2026, 6:29am

Assuming you are all experiencing the same issue (and I know we should be careful with assumptions) let’s try to find some common elements. @wyleu says this occurs with just HDMI and no touch, just VNC which would remove the touch interface from the equation. He also sees it only on a RPi4 machine. I have not seen this with a similarly configured RPi5. So can others say whether they are using RPi4 or RPi5?

Also, can others confirm whether they have 100% CPU on one core when this issue occurs? (Use top or htop in a ssh /terminal connection to see this.)

jawn · May 8, 2026, 7:04am

I had it happening on a V4 with Pi 4. If I see it again I’ll check the CPU.

smespresati · May 8, 2026, 7:10am

In my case I use rpi 5, adc2 dac pro and HDMI +USB. About CPU load I cannot check It right now as I’m away from home for a couple of weeks….but in my testing I think I saw CPU overload as well.