Skip to main content

​
Deployment Issues

Symptoms: Deployment status shows “failed” with a build error in the logs.Solutions:
  1. Check Settings > Deployment History and expand the failed deployment to read the error message
  2. Common causes:
    • Syntax errors in JavaScript/TypeScript files
    • Missing dependencies (run npm install in the terminal)
    • Invalid package.json or build script
  3. Fix the issue in the code editor and deploy again
  4. If the error is unclear, paste the error message into the AI chat in Code mode and ask it to find and fix the cause
A build error stops before anything reaches your live site, so visitors keep the version they had.
Symptoms: Deployment succeeded but the site shows nothing.Solutions:
  1. Open browser DevTools (F12) and check the Console tab for JavaScript errors
  2. Verify your index.html or entry point exists in the project files
  3. For React/Next.js apps, ensure the root component renders correctly
  4. Check the Hosting dashboard log viewer for server-side errors
  5. Ask the AI chat in Code mode: “My site shows a blank page after deploying. Find the cause and fix it.”
Symptoms: A monitor, a script or a curl from a terminal receives an HTTP 202 with an empty body, or a 403, while the site loads normally in a browser.Check the response headers. An automated client the firewall cannot identify gets a 202 challenge (empty body, x-amzn-waf-action: challenge) or a 403 with x-hiveku-firewall: blocked; a request from a known bulk-scraper network gets a 403 with x-hiveku-firewall: blocked-network; a 403 without that header comes from the site itself. Real visitors and real search engines are not affected.Solutions:
  1. 202 challenge or 403 with x-hiveku-firewall: blocked — allow the client under Hosting > Firewall. It is easiest to allow when it sends its own product name and version in its user agent (such as MyMonitor/2.1).
  2. 403 with x-hiveku-firewall: blocked-network — the client runs on a known bulk-scraper network. An allowance does not lift this block; run the client from another network.
  3. 403 without that header — your site refused the request itself (a protected page, an expired signed link, a missing file). Check the page.
  4. A monitor that only needs the status code can send a HEAD request, which is never challenged or blocked as an automated client (the rate limit and the network block still apply).
Read the status code and the header, not the text in the body. See Allow a monitor through the firewall.
Symptoms: A test that sends Googlebot or bingbot as its user agent, from an office, a laptop or a rented server, gets a 202 or a 403 instead of the page, or the Hosting > Firewall clients table shows a Googlebot or bingbot row.That test is refused on purpose: anyone can type those names. Real Googlebot and bingbot are verified by the address they come from and pass. To check what they actually see, use URL Inspection in Google Search Console and in Bing Webmaster Tools, and search the Hosting > Firewall clients table for Googlebot; a Googlebot row on Google Cloud (rented servers) is usually an impostor. A Blocked row is different, because Blocked also counts the 403s your own site sends: a Blocked Googlebot row on the Google network is usually your site’s own 403 to the real crawler (a protected page or a missing file), not the firewall, and a Blocked bingbot row on Microsoft / Azure can be too. Check that page, and confirm with URL Inspection. See Is Google or Bing blocked?.
Symptoms: Browser shows “Your connection is not private” or the page keeps redirecting.Solutions:
  1. DNS not propagated yet — wait 15-30 minutes after adding your CNAME record, then click Verify DNS Configuration in Settings
  2. Cloudflare users — set your CNAME proxy status to DNS only (gray cloud), not Proxied. If you must use Cloudflare proxy, set SSL/TLS mode to Full (strict)
  3. Check that your CNAME record points to the exact distribution address shown in the domain setup wizard
  4. Use dnschecker.org to verify DNS propagation globally

​
AI Assistant Issues

Symptoms: The AI won’t make changes and shows a credits warning.Solutions:
  1. Credits reset on the 1st of each month automatically
  2. Switch to Hiveku Mini (free on all plans) for basic tasks
  3. Ask your account Owner/Admin to upgrade the plan for more credits
  4. Connect a BYOK Anthropic API key in Settings > API & Connectors (where bring-your-own-key is turned on for your account) to bypass credits entirely
Symptoms: Messages take a long time or the stream hangs.Solutions:
  1. Try switching to a different model tier (Mini is fastest)
  2. Use /compact to condense a long conversation — this frees up context space
  3. Start a new session with /clear if the conversation has become very long
  4. Set reasoning effort to Low for simple tasks
  5. Check your internet connection
Symptoms: The AI edited files incorrectly or broke something.Solutions:
  1. In the chat, click Undo this turn above the request. It goes back to the version saved before the request, and keeps the newer versions so you can undo the undo. See Undo an AI turn.
  2. To go back further, open the Branches tab, pick the version that worked and click Roll back to this version. See Versions.
  3. For one file, open it in the code editor, click History, and Bring it back from an earlier copy.
Versions hold pages and code. If images, uploads or data need to come back too, use Project history.

​
Editor Issues

Symptoms: Changes in the editor don’t appear in the preview pane.Solutions:
  1. Make sure you saved the file (Ctrl+S / Cmd+S) — look for the dot indicator on the tab to disappear
  2. Try refreshing the preview manually using the refresh button in the preview toolbar
  3. For container-based projects, the hot reload may take a few seconds
  4. Check the browser console for JavaScript errors that might prevent rendering
Symptoms: Uploaded files don’t appear in the file tree.Solutions:
  1. Binary files (images, videos) are stored as assets — check the asset manager
  2. Make sure the file isn’t too large for your plan’s storage limit
  3. Try uploading again by dragging the file onto a specific folder in the file tree
  4. Check Settings > Resource Usage to verify you haven’t hit your storage limit

​
GitHub Issues

Symptoms: Committing or pushing changes shows an error.Solutions:
  1. Verify your GitHub connection in Settings > GitHub — reconnect if the token has expired
  2. Check that the Hiveku GitHub App still has access to the target repository
  3. If there are conflicts, pull the latest changes first, resolve conflicts, then push again
  4. Make sure the repository hasn’t been deleted or transferred on GitHub
Symptoms: Files you edited were replaced by the GitHub version.Solutions:
  1. A pull saves Your site as a version first (“Saved automatically before bringing in changes from GitHub”). Open the Branches tab, find that version, and click Roll back to this version.
  2. For one file, open the version and use Bring back this page.
A pull replaces your files with the GitHub version. The version Hiveku saves first is your way back.

​
Database Issues

Symptoms: Your app can’t connect to the database, or queries return errors.Solutions:
  1. Verify the database is provisioned in Project Settings > Database
  2. Check that your environment variables include the correct DATABASE_URL
  3. For BYOD (Bring Your Own Database), verify the connection string is correct and the database allows connections from Hiveku’s IP range
  4. Check the Hosting dashboard log viewer for connection error details
Go to Project Settings > Database > Backups and find the backup you want. Click Restore to replace the current database with that backup.
Restoring a backup replaces all current data. Save a restore point first in Project history (Save restore point) if you want to keep the current state.