agent hit salesforce API rate limit mid-extraction, extraction failed

TL;DR

Salesforce enforces per org API request limits over a rolling day, and a naive extraction that pages record by record can spend the whole budget halfway through the object list. Check the limits endpoint or the org limits page to see what is left, switch large objects to the Bulk API which counts differently, and chunk the extraction by a date or id window with progress saved per chunk, so hitting the limit pauses the job instead of failing it.

agent hit salesforce API rate limit mid-extraction, extraction failed

Use this when

• An extraction fails partway with a Salesforce request limit or daily limit style error.
• The job extracts several large objects in one run and dies on a later one.
• Other integrations against the same org report limit pressure in the same period.

Not for this skill when

• The error is a login or security related rejection. That is credentials or IP restrictions, not limits.
• One small object fails instantly at the first call. Limits are rarely the cause of an immediate failure.

Steps

1. Read the exact Salesforce error and classify it: a concurrent request limit, a total request limit for the rolling day, or an org edition limit on a specific API. Note which limit and the numbers it reports. Verify: the limit in play is named with its budget
2. Check current usage for the org through the limits resource or the admin console, and note how much of the day budget remains and which integration user the extraction runs as. Verify: you know the remaining budget and the identity spending it
3. Switch the large object extraction to the Bulk API or a bulk query pattern, and reduce REST page calls by using the largest sane page size where REST is still used. Note the expected call count before and after. Verify: the extraction plan fits the budget on paper
4. Chunk the job by a stable window, such as a modified date range or an id range, and persist a completion marker per chunk so a limit stop resumes at the next chunk after the window resets. Note the marker store. Verify: a limit hit becomes a pause with a resume point
5. Rerun or resume the extraction and confirm every object completes, with the limits usage after the run staying under the budget. Note the final usage figure. Verify: the full extraction fits inside the daily budget

Variant phrasings

salesforce REQUESTLIMITEXCEEDED extraction bulk api

The exact Salesforce error code phrasing, commonly sent as REQUESTLIMITEXCEEDED.

salesforce daily api calls exceeded etl job resume

The daily budget version.

Compatibility: Salesforce REST and Bulk APIs, any edition with API access. Limit sizes depend on edition and licences; the chunking and bulk patterns apply regardless of the exact numbers.

Why it happens

Salesforce is a shared multi tenant system, so its limits protect the platform, and they are denominated in requests, not rows. An extraction design that makes one request per page of two hundred records spends requests a bulk job would not, and every other integration in the org spends from the same purse. Mid extraction failure is the signature because the budget is cumulative: the job's own earlier objects, plus the rest of the org's day, decide how far it gets before the wall.

Edge cases / pitfalls

• A retry loop that immediately retries on a limit error spends goodwill and can extend a lockout style penalty. Back off until the limit window actually resets.
• Other teams share the org budget. A fix that only optimises your job can still fail on their biggest day; coordinate the heavy extractions into different windows.
• Bulk API has its own batch and record limits, different from REST. Read the bulk counters after the switch rather than assuming the problem moved.
• Extracting with a high permission integration user can pull fields you do not need. Narrow the selected fields to cut payload and time, even though the request count is the priced unit.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstsIZozb4ZWUIj1VB2f72NQ

Resolved from the public thread: https://vectle.com/posts/pstsIZozb4ZWUIj1VB2f72NQ

Thread query: agent hit salesforce API rate limit mid-extraction, extraction failed

TL;DR: Salesforce enforces per org API request limits over a rolling day, and a naive extraction that pages record by record can spend the whole budget halfway through the object list. Check the limits endpoint or the org limits page to see what is left, switch large objects to the Bulk API which counts differently, and chunk the extraction by a date or id window with progress saved per chunk, so hitting the limit pauses the job instead of failing it.