@sdetweil thanks for the idea. I already had this code seen. But never implemented in my pm2 startup file.
I tried up to 512 (limit of the raspberry I have).

Thats the log I got:
0|mm | <— Last few GCs —>
0|mm | [22565:0x47124000] 127560 ms: Mark-Compact (reduce) 127.2 (128.1) -> 126.7 (128.3) MB, pooled: 0 MB, 731.03 / 0.01 ms (+ 36.3 ms in 0 steps since start of marking, biggest step 0.0 ms, walltime since start of marking 896 ms) (average mu = 0.169, curren[22565:0x47124000] 128545 ms: Mark-Compact (reduce) 127.2 (128.3) -> 126.4 (127.8) MB, pooled: 0 MB, 832.78 / 0.00 ms (+ 1.0 ms in 0 steps since start of marking, biggest step 0.0 ms, walltime since start of marking 860 ms) (average mu = 0.162, current
0|mm | <— JS stacktrace —>
0|mm | FATAL ERROR: Ineffective mark-compacts near heap limit Allocation failed - JavaScript heap out of memory
0|mm | ----- Native stack trace -----
0|mm | /home/mirror/MagicMirror/node_modules/electron/dist/electron exited with signal SIGABRT

I rolled back to the old config before the change of the MMM-PublicTransprtHafas and it run hours (unfortunately I don’t get any live data anymore with the old version)…

More ideas? ;) Grateful for every hint :)

@sdetweil

while running and short time before crashing:
user:~ $ free -m
total used free shared buff/cache available
Mem: 426 217 99 19 109 132
Swap: 871 250 621

Hi, thanks for your update.
I installed your git version and I had to install undici additionally. Is that correct?

The MM works but ended starting up again and agin:
0|mm | <— Last few GCs —>
0|mm | [17870:0x2a530000] 119266 ms: Mark-Compact (reduce) 126.9 (128.6) -> 126.4 (128.1) MB, pooled: 0 MB, 791.61 / 0.00 ms (+ 0.9 ms in 0 steps since start of marking, biggest step 0.0 ms, walltime since start of marking 816 ms) (average mu = 0.190, current[17870:0x2a530000] 120521 ms: Mark-Compact 126.9 (128.1) -> 126.5 (128.6) MB, pooled: 0 MB, 1201.60 / 0.00 ms (average mu = 0.113, current mu = 0.043) allocation failure; scavenge might not succeed
0|mm | <— JS stacktrace —>
0|mm | FATAL ERROR: Ineffective mark-compacts near heap limit Allocation failed - JavaScript heap out of memory
0|mm | ----- Native stack trace -----
0|mm | /home/mirror/MagicMirror/node_modules/electron/dist/electron exited with signal SIGABRT

Im using a raspberry pi zero with 512 MB RAM which worked fine in the past.

Also some system information:
0|mm | [2025-02-03 08:57:53.254] [INFO] System information:
0|mm | ### SYSTEM: manufacturer: Raspberry Pi Foundation; model: Raspberry Pi 3 Model A Plus Rev 1.0; virtual: false
0|mm | ### OS: platform: linux; distro: Raspbian GNU/Linux; release: 11; arch: arm; kernel: 6.1.21-v7+
0|mm | ### VERSIONS: electron: 32.2.7; used node: 22.13.1; installed node: 22.13.1; npm: 10.9.2; pm2: 5.4.3
0|mm | ### OTHER: timeZone: Europe/Berlin; ELECTRON_ENABLE_GPU: undefined

Anyone have the same problem?

@karsten13

thanks again. The magic mirror now runs smoothly and without any errors :)

l@karsten13

hi, im running now in the recommended electron and a newer node version

SYSTEM: manufacturer: Raspberry Pi Foundation; model: Raspberry Pi 3 Model A Plus Rev 1.0; raspberry: [object Object]; virtual: false
OS: platform: linux; distro: Raspbian GNU/Linux; release: 11; arch: arm; kernel: 6.1.21-v7+
VERSIONS: electron: 31.1.0; used node: 20.14.0; installed node: 22.5.1; npm: 10.8.2; pm2: 5.4.1
OTHER: timeZone: Europe/Berlin; ELECTRON_ENABLE_GPU: undefined

The MM now starts very fast and without rebooting the raspberry. I’ll keep you updated.
Thanks in advance :)

@karsten13

here is the information:

[2024-07-22 07:54:19.741] [INFO] System information:

SYSTEM: manufacturer: Raspberry Pi Foundation; model: Raspberry Pi 3 Model A Plus Rev 1.0; raspberry: [object Object]; virtual: false

OS: platform: linux; distro: Raspbian GNU/Linux; release: 11; arch: arm; kernel: 6.1.21-v7+

VERSIONS: electron: 29.1.6; used node: 20.9.0; installed node: 21.7.3; npm: 10.5.0; pm2: 5.3.1

OTHER: timeZone: Europe/Berlin; ELECTRON_ENABLE_GPU: undefined

@sdetweil
I updated from 2.27.0 to 2.28.0 by script.
Sounds like 2.28.0 isn’t the newest release anymore? Or will it be fixed in the unreleased version?

What do you mean with the calendar entry? The config?

{
module: “calendar”,
header: “Müllkalender”,
position: “bottom_left”,
config: {
customEvents: [{keyword: ‘bio’, symbol: ‘recycle’, color: ‘green’},{keyword: ‘müll’, symbol: ‘trash’, color: ‘Green’}],
fetchInterval: 1200000,
fadePoint: 0.75,
calendars: [
{
symbol: “calendar-check”,
maximumEntries: 3,
url: “https://xxx”
}
]
}
},

thank you!

Hi,

after upgrading to the second community release Im stuck with these errors:

0|mm | [4143:0719/155732.710554:ERROR:gpu_process_host.cc(989)] GPU process exited unexpectedly: exit_code=512

0|mm | [2024-07-19 15:58:46.475] [ERROR] Calendar Error. Could not fetch calendar: MY-iCLOUD-CALENDAR-ADRESS TypeError: fetch failed
0|mm | at Object.fetch (node:internal/deps/undici/undici:12293:11)
0|mm | at processTicksAndRejections (node:internal/process/task_queues:95:5)
0|mm | at runNextTicks (node:internal/process/task_queues:64:3)
0|mm | at process.processImmediate (node:internal/timers:449:9) {
0|mm | cause: ConnectTimeoutError: Connect Timeout Error
0|mm | at onConnectTimeout (node:internal/deps/undici/undici:7434:28)
0|mm | at node:internal/deps/undici/undici:7390:50
0|mm | at Immediate._onImmediate (node:internal/deps/undici/undici:7422:13)
0|mm | at process.processImmediate (node:internal/timers:478:21) {
0|mm | code: ‘UND_ERR_CONNECT_TIMEOUT’
0|mm | }

Im running on a raspberry pi 2B+ with the node version v21.7.3

Any ideas or known problems? I don’t find any in combination of these two errors.
Thanks in advance :)