Grok Bot 0.30.0 unusable for 4+ days — can't reach computer / stuck previous image, webhook routines still burn usage

Support Bug Reports
,
1

Where does the bug appear (feature/product)?

Grok Bot

Describe the Bug

Grok Bot has been completely unusable for the past 4+ days on my account.

The hosted Agent Computer never comes online. I keep hitting connection / reachability failures, and when I try to update or recover the computer I get:

the update cannot be completed as the agent’s computer remains on its previous image

Reset and Restart both fail. This is not limited to one bot or one conversation — the entire shared computer is down, so every bot on the account is unusable.

There is a second product bug on top of the offline computer: webhook / routine jobs still fire while the computer is unreachable. I cannot pause or disable those routines myself because:

  • Opening Routine controls appears to require the computer to be active
  • Messaging the bot to pause stays stuck on “waiting for the bot and computer to come online”

So the computer is dead for interactive use, but background routines keep running.

This looks like the same backend stuck Agent Computer state other users reported (local Reset/Recover/reinstall often cannot clear it and a server-side rebuild is needed). Staff has also asked people not to pile onto closed megathreads and to open distinct reports, so this is my account-specific report.

Steps to Reproduce

  1. Open Grok Bot desktop (v0.30.0).
  2. Wait for bots / shared computer to load.
  3. Observe persistent unreachable / reconnect failure (computer never becomes reachable).
  4. Attempt Update Agent Computer → fails with “update cannot be completed as the agent’s computer remains on its previous image”.
  5. Attempt Recover / Reset / Restart → fail or partial state (“Reset failed. Your computer is in a partial state…”).
  6. Fully quit (Cmd+Q), reopen, retry reconnect once → same failure.
  7. Try to pause webhook/routine runs from agent details or by messaging the bot → blocked because computer is offline; messages stay “waiting for the bot and computer to come online”.
  8. Observe that webhook/routine activity continues even though no interactive session works.

Expected Behavior

  1. The hosted Grok Bot Agent Computer should connect and load existing bots, files, and sessions.
  2. Update / Recover / Reset should either succeed or clearly complete a backend rebuild without leaving a permanent “previous image” / partial stuck state.
  3. If the computer is down, scheduled/webhook routines should stop, or there must be a working offline path to pause/disable all routines.
  4. Interactive Grok Bot should be usable again after a backend rebuild if local recovery cannot clear the stuck state.

Screenshots / Screen Recordings

grokbot-error-screenshot.png
grokbot-error-screenshot.png3824×2254 345 KB

Operating System

MacOS

Version Information

Grok Bot:
Version: 0.30.0
Release Track: stable
OS: darwin / macOS

For AI issues: which model did you use?

N/A — Grok Bot computer / connectivity issue. Product is Grok Bot itself.

For AI issues: add Request ID with privacy disabled

N/A — not a single-chat model failure. Entire Agent Computer is unreachable for all bots. Happy to provide Conversation IDs if useful after the box is restored.

Additional Information

What I already tried (multiple times):

  • Fully quit and reopen Grok Bot (Cmd+Q)
  • Reconnect / Retry
  • Recover Agent Computer (once, waited)
  • Reset Agent Computer (fails / partial state)
  • Restart
  • Update Agent Computer (fails: remains on previous image)
  • Attempted to pause webhook routines from UI and via bot message (blocked while offline)

Related existing reports (same/similar categories — titles only, not asking to reopen closed megathreads):

Connectivity / can’t reach computer (tags: grok-bot, networking):

  • Grok Bot: “Can’t reach your computer” for 48+ hours
  • Grok Bot 0.30.0 — All bots unavailable — “Can’t reach your computer”
  • Grok Bot Computer Stuck in Partial State — Reset Failed, Existing Cloud Computer Unreachable
  • Grok Bot can’t reach computer — white loading dot
  • Grok Bot Windows — fresh profile setup fails with “Can’t reach your computer” after backend fix

Routines continuing while user cannot control the bot:

  • Grok Bot: weekly usage hits 100% after bot-to-bot reviews the user asked to stop

Ask:
Please rebuild the hosted Agent Computer for my account on the backend, and stop or pause webhook/routine runs while the box is down. This is a Grok Bot connectivity / computer-stuck bug, not a billing request.

This thread was auto-moved to Account & Billing and unlisted. Requesting it be moved back to Bug Reports and re-listed — primary issue is “Can’t reach computer” / stuck previous image / Reset failed partial state.

cc @deanrie @Colin

Does this stop you from using Cursor

Sometimes - I can sometimes use Cursor

(Cursor IDE itself still works. Grok Bot is fully unusable.)

4

Hi there!

Thanks for reaching out. This looks like a billing-related query, and the forum isn’t the right channel for these. The forum team does not have access to account details or the ability to do things like process refunds or manage subscriptions. As such, we’ve unlisted this thread.

For anything specific to your account, or if you’re unsure, please email [email protected]. Our billing specialists are the experts in this domain and are best equipped to handle your request.

If you’ve already emailed, hang tight. The team will get back to you as soon as they can!

5

Update / root cause found

I dug further and this looks like a DNS / ISP resolver issue, not a stuck Agent Computer on Cursor’s backend.

What I found

  • ISP: Reliant Geo Infocom (India)
  • Their DNS appears to be blocking / failing to resolve Grok Bot’s Cursor VM hostnames (the *.cursorvm.com / private agent-computer addresses)
  • Cursor IDE / other Cursor APIs could still look fine, but Grok Bot could not reach the cloud computer
  • Symptoms matched “Can’t reach your computer”, Reset/Recover failures, and the partial-state UI

What fixed it

I changed DNS away from the ISP resolver to Cloudflare DNS (1.1.1.1 / 1.0.0.1).

After that:

  • Grok Bot connected normally again
  • The agent computer became reachable
  • Things started working

So the earlier “stuck previous image / Reset failed” state was likely a side effect of the client never being able to resolve/reach the VM endpoints via ISP DNS.

Questions for the team

  1. Remediation on Cursor’s side?
    Is there anything planned so this fails more clearly (e.g. “DNS cannot resolve cursorvm” instead of generic can’t-reach / reset-failed UI)? Or any in-app network check that flags bad ISP DNS for *.cursorvm.com?

  2. Will this keep happening?
    If I stay on Reliant Geo Infocom DNS, should I expect this to break again whenever their resolver filters or breaks those records? Is the durable fix just:

    • keep device/router DNS on Cloudflare (1.1.1.1) or Google (8.8.8.8)
    • set both IPv4 and IPv6 DNS (so router RA doesn’t silently fall back to ISP DNS)
    • optionally use DNS-over-HTTPS so the ISP can’t rewrite plain DNS

  3. Any recommended permanent setup for users in India / similar ISPs?

Happy to share resolver details / nslookup output if useful.

cc @deanrie @Colin

Also: this thread was auto-moved to Account & Billing and unlisted, but the root cause is networking/DNS. Please move back to Bug Reports / re-list if possible.

8

Hey, great diagnosis, your breakdown is spot on. What you’re describing matches a pattern we’ve been tracking: the Grok Bot desktop client can’t resolve the agent machine hostnames (*.cursorvm.com), while the box itself is actually online. Some ISP DNS resolvers, especially in India, don’t return these records. From the outside, it looks like “Can’t reach your computer” or “stuck on previous image” or Reset failed, but the real cause is DNS. Switching to Cloudflare DNS 1.1.1.1 fixes this specific issue.

On your questions:

  1. A clearer error message or a built-in DNS check: agreed, that’s the right direction. The generic “can’t reach” is misleading in this case. I passed this to the team as feedback. I can’t share an ETA.

  2. Yes, if you switch back to an ISP resolver that filters these records, it can happen again. Your plan for a reliable fix is correct:

  • Set DNS on your router or device to Cloudflare 1.1.1.1 and 1.0.0.1 or Google 8.8.8.8 and 8.8.4.4
  • Set both IPv4 and IPv6 DNS, so router RA doesn’t quietly fall back to the ISP resolver
  • DNS-over-HTTPS is optional if your ISP rewrites plain DNS
  1. For India or similar ISPs, that’s the recommended setup: a public resolver Cloudflare or Google on both stacks, plus optional DoH.

About the usage that was billed while the box was unreachable, that’s account-specific. Best to reach out to the team directly at [email protected] and they can check your account.

I moved the thread back to Bug Reports and re-listed it, the root cause here is network-related. If you still hit unreachable after switching to Cloudflare, reply here with your nslookup output for the machine hostname and we’ll take a look.