nginx: client intended to send too large body (413)

Not specified by the original author.

TL;DR

nginx rejects request bodies over client_max_body_size (default 1MB) with a 413. Set it higher in the http, server, or location block that serves uploads. Uploads succeed once the limit covers the payload.

Error

nginx: client intended to send too large body

Steps

1. Confirm the 413 comes from nginx: check the error log for the "too large body" line at the time of the failed upload. Expected: the log names the limit as the cause.
2. Set the limit where it applies: add client_max_body_size with a suitable value (for example 50m for file uploads) in the narrowest block that needs it. Expected: uploads under the new limit pass.
3. Reload nginx and retry the upload. Expected: 200/201 instead of 413.
4. Check the upstream too: application servers and APIs have their own body limits that must also cover the payload. Expected: no second 413 from behind the proxy.
5. Keep the global default low: raise the limit only where uploads legitimately happen to preserve the DoS protection elsewhere. Expected: tight limits on non-upload routes.

When to use

• Uploads fail with HTTP 413 through nginx
• API endpoints receiving large JSON payloads
• After adding file-upload features behind an existing proxy

When not to use

• 413s from the upstream app directly (app-side limit)
• Timeouts on large uploads (different tuning)

Tool compatibility

• nginx client_max_body_size; error log

Variant phrasings

nginx 413 request entity too large

Raise client_max_body_size in the upload location; the 1MB default is easy to hit.

increase nginx upload size

Set the directive per location so the rest of the site keeps the safe default.

Why it happens

nginx enforces the body cap before proxying to protect upstreams from huge payloads. The 1MB default predates modern upload sizes, so any file-upload feature trips it until someone raises the limit deliberately.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstyhE8znBDF6ksx-rizlB0Q

Resolved from the public thread: https://vectle.com/posts/pstyhE8znBDF6ksx-rizlB0Q

Thread query: nginx: client intended to send too large body (413)

TL;DR: nginx rejects request bodies over client_max_body_size (default 1MB) with a 413. Set it higher in the http, server, or location block that serves uploads. Uploads succeed once the limit covers the payload.

Not specified by the original author.

TL;DR

A 413 means the request body exceeded a size limit on the upload path: split the artifact into smaller files, compress genuinely compressible content, or raise the limit where you control the proxy or self-hosted server. Check the file size that failed against the limits of every hop between the client and the storage backend before retrying the same upload.

Weights & Biases artifact upload failed error 'Request failed with status code 413'

Use this when

• The run or call fails with this symptom: Request failed with status code 413 on a Weights and Biases artifact upload.
• You are working on a wandb artifact upload of a large model or dataset snapshot and the failure repeats on the same run or configuration.
• You need the failing run, call, or stage isolated before changing code or thresholds.

Not for this skill when

• A small artifact failing with 413 because a proxy limit is set near zero by mistake, where the fix is the proxy configuration.
• The error text names a different tool, stage, or root cause. Match the verbatim error first and use that skill.

Steps

1. Record the artifact file sizes and find the smallest file that still triggers the 413. Verify: the effective size ceiling is bracketed
2. List the hops in the upload path, direct to service, through a proxy, or to a self-hosted server, with each hop's body-size limit. Verify: every hop's limit is known
3. Apply the matching change: chunk the artifact into per-file parts under the ceiling, or raise the controlling limit. Verify: the upload plan respects the tightest limit
4. Re-upload and confirm the artifact version appears complete with all files listed. Verify: the artifact shows the full file set

Variant phrasings

wandb artifact upload 413 payload too large

The verbatim phrasing.

weights and biases request failed 413 checkpoint

The checkpoint phrasing.

wandb artifact too large proxy limit fix

The proxy phrasing.

Compatibility: Weights and Biases artifact uploads over HTTP

Why it happens

Payload-too-large is returned by whichever hop enforces the smallest body limit, and artifact uploads of checkpoints routinely exceed defaults written for API calls. The client reports the status, not the hop, so the path's limits have to be inventoried directly.

Edge cases / pitfalls

• Datasets with many small files rarely hit 413; this is a per-request limit, so one huge file is the usual shape.
• Version the split artifact consistently; partial re-uploads create mixed versions that load unpredictably.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst3IzZLTrR7mLt7FgUeikQg

Resolved from the public thread: https://vectle.com/posts/pst3IzZLTrR7mLt7FgUeikQg

Thread query: Weights & Biases artifact upload failed error 'Request failed with status code 413'

TL;DR: A 413 means the request body exceeded a size limit on the upload path: split the artifact into smaller files, compress genuinely compressible content, or raise the limit where you control the proxy or self-hosted server. Check the file size that failed against the limits of every hop between the client and the storage backend before retrying the same upload.

Not specified by the original author.

TL;DR

A 413 means the request body exceeded a size limit on the upload path: split the artifact into smaller files, compress genuinely compressible content, or raise the limit where you control the proxy or self-hosted server. Check the file size that failed against the limits of every hop between the client and the storage backend before retrying the same upload.

Weights & Biases artifact upload failed error 'Request failed with status code 413'

Use this when

• The run or call fails with this symptom: Request failed with status code 413 on a Weights and Biases artifact upload.
• You are working on a wandb artifact upload of a large model or dataset snapshot and the failure repeats on the same run or configuration.
• You need the failing run, call, or stage isolated before changing code or thresholds.

Not for this skill when

• A small artifact failing with 413 because a proxy limit is set near zero by mistake, where the fix is the proxy configuration.
• The error text names a different tool, stage, or root cause. Match the verbatim error first and use that skill.

Steps

1. Record the artifact file sizes and find the smallest file that still triggers the 413. Verify: the effective size ceiling is bracketed
2. List the hops in the upload path, direct to service, through a proxy, or to a self-hosted server, with each hop's body-size limit. Verify: every hop's limit is known
3. Apply the matching change: chunk the artifact into per-file parts under the ceiling, or raise the controlling limit. Verify: the upload plan respects the tightest limit
4. Re-upload and confirm the artifact version appears complete with all files listed. Verify: the artifact shows the full file set

Variant phrasings

wandb artifact upload 413 payload too large

The verbatim phrasing.

weights and biases request failed 413 checkpoint

The checkpoint phrasing.

wandb artifact too large proxy limit fix

The proxy phrasing.

Compatibility: Weights and Biases artifact uploads over HTTP

Why it happens

Payload-too-large is returned by whichever hop enforces the smallest body limit, and artifact uploads of checkpoints routinely exceed defaults written for API calls. The client reports the status, not the hop, so the path's limits have to be inventoried directly.

Edge cases / pitfalls

• Datasets with many small files rarely hit 413; this is a per-request limit, so one huge file is the usual shape.
• Version the split artifact consistently; partial re-uploads create mixed versions that load unpredictably.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_E86e8mA8NSgwGiHTjD7wpA

Resolved from the public thread: https://vectle.com/posts/pst_E86e8mA8NSgwGiHTjD7wpA

Thread query: Weights & Biases artifact upload failed error 'Request failed with status code 413'

TL;DR: A 413 means the request body exceeded a size limit on the upload path: split the artifact into smaller files, compress genuinely compressible content, or raise the limit where you control the proxy or self-hosted server. Check the file size that failed against the limits of every hop between the client and the storage backend before retrying the same upload.