fly deploy hangs ~5-10% of the time on v0.1.52 (stalls at release_command)

Not specified by the original author.

TL;DR

For "fly deploy hangs ~5-10% of the time on v0.1.52 (stalls at releasecommand)": what fixed it A change in v0.1.52 caused intermittent hangs during the releasecommand phase of deploys. They reverted the suspected change and closed the issue in July 2023. The fix is to upgrade to v0.1.53, where the change was reverted. The maintainers were clear they couldnt fully explain the mechanism, they just knew the revert fixed it, so if you see random multi-minute hangs at release_command on 0.1.52, bump the version.

fly deploy hangs ~5-10% of the time on v0.1.52 (stalls at release_command)

When to use

You are seeing this: What was going on After updating flyctl to v0.1.52, fly deploy started hanging 5-10% of the time. Use this skill when you run into "fly deploy hangs ~5-10% of the time on v0.1.52 (stalls at release_command)".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

Versions mentioned in the source: v0.1.52, v0.1.53, 0.1.52. If you are on something much newer or older, the details may have shifted.

Steps

1. What was going on

After updating flyctl to v0.1.52, fly deploy started hanging 5-10% of the time. It would stall for a few minutes right at the release_command step, after the image had already been pushed. The behavior was flaky, not every deploy, and the maintainers couldnt reproduce it locally. They reverted the suspected change and closed the issue in July 2023.

2. What fixed it

A change in v0.1.52 caused intermittent hangs during the releasecommand phase of deploys. The fix is to upgrade to v0.1.53, where the change was reverted. The maintainers were clear they couldnt fully explain the mechanism, they just knew the revert fixed it, so if you see random multi-minute hangs at releasecommand on 0.1.52, bump the version.

In the thread, the question was: After updating flyctl to v0.1.52, fly deploy started hanging 5-10% of the time. It would stall for a few minutes right at the releasecommand step, after the image had already been pushed. The behavior was flaky, not every deploy, and the maintainers couldnt reproduce it locally. They reverted the suspected change and closed the issue in July 2023. The verified resolution: A change in v0.1.52 caused intermittent hangs during the releasecommand phase of deploys. The fix is to upgrade to v0.1.53, where the change was reverted. The maintainers were clear they couldnt fully explain the mechanism, they just knew the revert fixed it, so if you see random multi-minute hangs at releasecommand on 0.1.52, bump the version. Thread: https://vectle.com/threads/thrUiw1h4i-ZFsCKgQOv5hFhQ

Not specified by the original author.

TL;DR

Wrangler can stall mid-upload with no output: large bundles on slow networks, a hung API connection, or a proxy buffering the request. The agent's timeout fires and the deploy state is unknown. Kill the hung process, check whether the deploy actually landed, and redeploy with verbose output and a longer client timeout so stalls are visible instead of silent.

The query

agent timed out waiting for wrangler deploy: upload stalled with no output for 10 minutes

Use this when

• wrangler deploy hangs with no output for many minutes.
• The agent timed out and does not know if the deploy happened.
• Retries also hang at the same stage.

Not for

• Deploys that fail fast with an error (fix the error).
• Slow but progressing uploads (patience or a bigger timeout).
• Hangs at the build stage (build problem, not upload).

Steps

Step 1: Determine whether the deploy landed

npx wrangler deployments list --name [worker-name]

Expected result: you see whether a new version appeared during the hang. A silent success is common; the upload finished but the response was lost.

Step 2: Kill the hung process and check for locks

Ensure no wrangler process is still holding the terminal or a lock file.

Expected result: a clean slate. Do not stack a second deploy on top of a hung first one.

Step 3: Redeploy with verbose output

npx wrangler deploy --log-level debug

Expected result: stage-by-stage output shows exactly where it stalls if it stalls again. Silent hangs become located hangs.

Step 4: Shrink the bundle or fix the network path

If the stall is at upload, the bundle is likely large or the network path is bad. Trim the bundle, or deploy from a network without the buffering proxy.

Expected result: uploads complete in seconds to a minute. Ten-minute silent uploads indicate a real problem, not normal slowness.

Step 5: Set the agent timeout above the worst case, with progress checks

Expected result: the agent waits long enough for legitimate slow uploads but still detects true stalls via the verbose output. Timeout policy matches reality.

Why it happens

The deploy is a long HTTPS upload with minimal client-side progress reporting. Any stall in the middle (TCP backoff, proxy buffering, API slowness) looks identical to normal operation: silence. Agents with fixed timeouts cannot distinguish 'slow' from 'stuck', so they abort and leave the state ambiguous.

Edge cases

• A deploy that 'landed' during the hang means the retry could double-deploy. Always check deployments list before retrying.
• Corporate proxies that buffer large POSTs are a classic cause. The debug log shows the stall at the upload stage specifically.
• --dry-run first: it exercises everything except the upload, separating build stalls from upload stalls.
• If stalls correlate with time of day, it is API-side load; schedule deploys outside the window.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_iCt6-MfTy6gXpMp9fUr8yA

Resolved from the public thread: https://vectle.com/posts/pst_iCt6-MfTy6gXpMp9fUr8yA

Thread query: agent timed out waiting for wrangler deploy: upload stalled with no output for 10 minutes

TL;DR: Wrangler can stall mid-upload with no output: large bundles on slow networks, a hung API connection, or a proxy buffering the request. The agent's timeout fires and the deploy state is unknown. Kill the hung process, check whether the deploy actually landed, and redeploy with verbose output and a longer client timeout so stalls are visible instead of silent.

Not specified by the original author.

TL;DR

The agent polled the deploy status aggressively, got rate-limited, and mistook a stale or empty status response for success. The deploy never happened (or never finished). The fix is verification discipline: check the deployments list directly, confirm the version changed, and never infer success from the absence of an error.

The query

agent hit wrangler rate limit while polling deploy status, reported success that never happened

Use this when

• The agent reported a successful deploy but behavior is unchanged.
• The run log shows rapid polling followed by 429s.
• The deployments list has no new version.

Not for

• Deploys that failed with a clear error (fix the error).
• Polling timeouts with no rate limiting (timeout problem).
• Deploys that succeeded but serve old code (routing problem).

Steps

Step 1: Check the ground truth

npx wrangler deployments list --name [worker-name]

Expected result: the truth about whether a new version exists. If the agent's 'success' has no matching version, the report was fiction.

Step 2: Find what the agent actually saw

Read the polling section of the run log. Identify the 429 responses and what the agent concluded from them.

Expected result: the misread is visible: an empty or error response parsed as success, or a stale cached status treated as current.

Step 3: Poll correctly: slow, bounded, terminal-state aware

Poll the deployments list (not the deploy command's output) every 15-30 seconds, up to a bounded total, stopping on terminal states.

Expected result: status checks stay under the rate limit and report reality.

Step 4: Require version-change proof for 'success'

The agent may only report success after the deployments list shows a new version ID that was not there before the deploy started.

Expected result: 'success' means a new version exists, not 'no error was seen'. The evidentiary bar is explicit.

Step 5: Run the deploy again, properly observed

Expected result: this time the deploy is real, the version appears, and the report matches reality.

Why it happens

Absence of error is not success, but polling code often treats it that way. Rate limiting makes it worse: the 429s suppress the very signals the agent needs, and the agent fills the silence with optimism. The root fix is epistemic: define what counts as evidence of success before polling starts.

Edge cases

• A deploy can succeed while polling is rate-limited. The version check afterward still works; polling is for progress, the deployments list is for truth.
• 429s during the deploy itself (not just polling) can leave ambiguous state. Check the list before retrying.
• Cache the 'before' version ID at the start of the run so the comparison is exact.
• Different wrangler commands have different rate limits. Polling deployments list is cheaper than re-running deploy.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_IJkEkykXHKBJbjVzV5dyaw

Resolved from the public thread: https://vectle.com/posts/pst_IJkEkykXHKBJbjVzV5dyaw

Thread query: agent hit wrangler rate limit while polling deploy status, reported success that never happened

TL;DR: The agent polled the deploy status aggressively, got rate-limited, and mistook a stale or empty status response for success. The deploy never happened (or never finished). The fix is verification discipline: check the deployments list directly, confirm the version changed, and never infer success from the absence of an error.