Grokbot desktop app crashs 0.78 - 1.0.1

Where does the bug appear (feature/product)?

Grok Bot

Describe the Bug

Grok Bot desktop app 1.0.1 keeps crashing with SIGTRAP on Ubuntu 26.04

Steps to Reproduce

It happens at random times, and nothing specific seems to set it off.

My guess is a memory overflow, todays crash started soon after boot. But it is random, times im doing nothing others it’s during load.

Expected Behavior

Not crash

Operating System

Linux

Version Information

Grok Bot 1.0.1 (amd64)

Additional Information

  • Ubuntu 26.04, kernel 7.0.0-43-generic

  • Lenovo Legion 5 (15AHP10), Ryzen 7 260, Radeon 780M plus RTX 5060 (hybrid graphics)

Ubuntu’s crash handler saved /var/crash/_opt_Grok Bot_grok-bot.1000.crash

I can share the full crash file if a developer needs it. It’s about 42 MB.

  • 16 GB RAM, about 10 GB free at the time, with no out-of-memory kill

Does this stop you from using Cursor

No - Cursor works, but with this issue

Hi @bee,

SIGTRAP usually means one of the app’s processes stopped itself after an internal check failed (that can include hitting its own memory limit), so it wouldn’t show up as an out-of-memory kill even with RAM free.

Please keep the crash file, but don’t post it here since it includes a memory dump. One thing would narrow it down a lot: which of the app’s processes crashed. Could you run this and paste just the --type=... part of the output, or say if there isn’t one?

grep -a -m1 '^ProcCmdline' "/var/crash/_opt_Grok Bot_grok-bot.1000.crash"

There’s no --type= in it. ProcCmdline is just /opt/Grok\ Bot/grok-bot\ --, so it’s the main process.

One more thing would help pin down where it stops. Right after the next crash (or now, if you haven’t rebooted since the last one), could you run this and paste the output? Skim it first for anything personal.

journalctl -b --no-pager | grep -i "grok-bot" | tail -n 40

i ran it just to check, nothing seems to point towards a crash reason just some minor camera things and small blip of connection issues with the chat stream.

ill save that command to run next time the crash happens to see if i can scoop it up.

Nov 12 09:09:31 Levo grok-bot.desktop[12020]: (node:12020) ExperimentalWarning: vm.USE_MAIN_CONTEXT_DEFAULT_LOADER is an experimental feature and might change at any time
Nov 12 09:09:31 Levo grok-bot.desktop[12020]: (Use grok-bot --trace-warnings ... to show where the warning was created)
Nov 12 09:09:31 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail pinned ac07b310-a547-4c02-815e-15e26e9d711a
Nov 12 09:09:31 Levo grok-bot.desktop[12334]: [12334:1112/090931.483011:ERROR:services/video_capture/video_capture_service_impl.cc:202] Bind context provider failed.
Nov 12 09:09:47 Levo grok-bot.desktop[12020]: [sand] link preview without image: page-http-4xx
Nov 12 09:12:24 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail dropped (attempt 1): ConnectError (ECONNRESET)
Nov 12 09:20:21 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail evicted 3 settled agent state(s) holding 813 row(s); 1 state(s) holding 519 row(s) retained
Nov 12 09:54:32 Levo grok-bot.desktop[244483]: [244483:1112/095432.668957:ERROR:services/video_capture/video_capture_service_impl.cc:202] Bind context provider failed.
Nov 12 10:05:34 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail evicted 1 settled agent state(s) holding 198 row(s); 1 state(s) holding 602 row(s) retained
Nov 12 10:28:30 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail evicted 1 settled agent state(s) holding 114 row(s); 2 state(s) holding 814 row(s) retained
Nov 12 10:31:41 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail evicted 1 settled agent state(s) holding 201 row(s); 1 state(s) holding 622 row(s) retained
Nov 12 11:14:17 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail evicted 1 settled agent state(s) holding 114 row(s); 1 state(s) holding 684 row(s) retained
Nov 12 11:16:12 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail dropped (attempt 1): ConnectError (ECONNRESET)
Nov 12 11:21:09 Levo grok-bot.desktop[12020]: sand.server_roster_read issued=50 joined=18 cached=20
Nov 12 11:24:37 Levo grok-bot.desktop[12268]: node-agent-coordinator: server transcript tail dropped (attempt 1): ConnectError (ECONNRESET)

Hey so i found the issue, and i believe it is the same type of issue with grokbot cause the issue happens similarly with Cursor the bash you gave me only works with current session reports so the “-1” was actually needed cause crash isn’t happening in the current session, its during shut down and the report pops up the next boot session.

Which explains why i kept getting pop ups but the program would not crash/close itself and why reports kept having the wrong date/time. But it’s the same SIGTRAP the grokbot app was having, same previous session crash pop up style, same depends.

Cursor crashes with SIGTRAP (“GPU process isn’t usable. Goodbye.”) when the user session ends on Linux

  • Cursor: 3.30.14 (deb, /usr/share/cursor/cursor)
  • OS: Ubuntu 26.04, kernel 7.0.0-43-generic, GNOME on Wayland
  • Hardware: AMD Ryzen 7 260 with Radeon 780M, plus an NVIDIA RTX 5060 (hybrid graphics)

What happens: when the GNOME session logs out or shuts down while Cursor is running, Cursor crashes instead of exiting cleanly. Ubuntu saves a crash report and shows the “cursor crashed” popup on the next login.

**Log, from journalctl -b -1:

**
cursor.desktop: ERROR:content/browser/network_service_instance_impl.cc:721] Network service crashed or was terminated, restarting service.

cursor.desktop: ERROR:content/common/zygote/zygote_communication_linux.cc:291] Failed to send GetTerminationStatus message to zygote

cursor.desktop: ERROR:content/browser/gpu/gpu_process_host.cc:1102] GPU process launch failed: error_code=1002 (x6)

cursor.desktop: FATAL:content/browser/gpu/gpu_data_manager_impl_private.cc:416] GPU process isn't usable. Goodbye.**

Crash report:** SignalName SIGTRAP, with the CPU stopped on a ud2 instruction. The stack has no symbols. RSS was about 375 MB, so it wasn’t out of memory.

What I think is going on: during session teardown, the network service and zygote helper processes are killed first. Cursor then tries to relaunch the GPU process, can’t, and hits the fatal CHECK instead of quitting quietly. Expected behavior is a clean exit on SIGTERM.**

**