Grok Bot cloud computer frozen after repeated snapshot rollbacks and Reset — all commands hang, recent work wiped (T-I62634)

Support Bug Reports
, ,
1

Where does the bug appear (feature/product)?

Grok Bot

Describe the Bug

The shared Grok Bot cloud Linux computer has been unstable for days and is now completely unresponsive.
On 30 Oct 2026 it spontaneously switched to older snapshots at least five times (known times ~12:47 PM, 1:38 PM, 2:52 PM MT). It alternated between a 125 GB disk and an older 128 GB snapshot. Each switch:
• Wiped that day’s files
• Restored previously deleted folders (e.g. /workspace/pkgsrc, old git worktrees)
• Spiked disk and inode usage (hit 100 % inodes at one point; later 67 % disk / 92 % inodes)
“Update Bot’s Computer” hung twice (~4:39 PM and 5:29 PM MT) and was blocked with “can’t update until you’re finished” even after work was stopped. After “Reset Bot’s Computer” at ~5:45 PM MT, every command hangs indefinitely. Even timeout 10 uptime; df -h / never returns. Calls hang 16–155 seconds then get cut off with messages such as:

Shell command was interrupted by the user after 155146ms before it completed.

No Request IDs or error codes are returned. In-app feedback fails with “Feedback could not be sent (unavailable).”
~20 bots are affected. Support ticket T-I62634 was emailed to [email protected] (urgent follow-up 29 Oct). Roughly $800 of paid usage this month is unusable. Work is safely backed up to Drive and Mac, but development is fully blocked.

Steps to Reproduce

  1. Use Grok Bot (desktop or web) with multiple agents on the shared computer.
  2. Observe spontaneous snapshot rollbacks that restore deleted files and wipe recent work.
  3. Try “Update Bot’s Computer” — it hangs or is blocked.
  4. Perform “Reset Bot’s Computer”.
  5. Attempt any shell command or file operation — it hangs indefinitely with no response or Request ID.

Expected Behavior

The cloud computer should stay stable, not spontaneously roll back snapshots, allow Update/Reset to complete, and respond promptly to commands so paid usage can be used.

Operating System

Windows 10/11
MacOS
iOS

Version Information

Grok Bot app version: unknown (check Update Status in the app). Cloud computer is Linux. No Request IDs returned on the failed calls.

For AI issues: which model did you use?

N/A — this is not a model/AI issue; it is the shared cloud computer infrastructure.

For AI issues: add Request ID with privacy disabled

None returned. Failed commands hang and are cut off without generating a Request ID.

Additional Information

Support ticket: T-I62634 (emailed from [email protected], follow-up 29 Oct 1:56 PM MT)
• Exact hang message: “Shell command was interrupted by the user after 155146ms before it completed.” (similar interruptions at ~16 s, 95 s, 100 s)
• In-app feedback: “Feedback could not be sent (unavailable).”
• Update blocked with “can’t update until you’re finished” / “PC Game Builder is still going”
• ~20 bots affected; disk/inode spikes during the restore cycles
• Happy to provide more logs, screenshots of the Update hang, or access if needed. Escalation requested.

Does this stop you from using Cursor

Yes - Cursor is unusable

5

Hey there, and sorry about this. What you’re describing isn’t intended behavior, and it isn’t anything you misconfigured. It’s an issue we’re tracking.

Most important for now: any time the computer looks rolled back or like saving has failed, hold off on Reset, Update, and Recover. Those rebuild from the last saved copy, so they can bring the old state back and clear newer files instead of fixing it. Safe order to try things is here: Grok Bot computer recovery. Keeping your own backups on Drive and your Mac was exactly the right call.

Could you fully quit Grok Bot (on Mac, Quit from the menu bar, not just closing the window) and reopen it, then let me know if it responds again? That’ll tell us whether you’re unblocked.

For anything specific to your account, your support email is the best place, and we’ll keep following up there.

6

@mohit Thanks for the reply. I haven’t heard anything from support after 5 days or so, but it’s been 24 hours since the last box reset. Grok bot was completely unresponsive for about 6 hours due to the flooding resets, unable to connect to a computer, unable to chat. It seems like a human may have fixed it on the backend, there hasn’t been a box reset now for 2 days. Grok bot seems to be functioning normally. I’m not positive it’s been solved, there’s been no feedback to my support ticket, but it does seem like it’s working again. It was happening about once per hour of usage, sometimes more frequently.