API limits and what happens when you pass them

Zendesk limits how many API calls you can make per minute. Passing the limit does not fail loudly, which is the problem.

Zendesk limits how many API calls you can make per minute. Passing the limit does not fail loudly, which is the problem.

How the limit works

The Help Center API shares that allowance rather than having its own. Go over and the API answers 429 Too Many Requests. Two headers come back with it: Retry-After says how many seconds to wait, and X-Rate-Limit reports your account's limit.

A well-written integration waits the Retry-After interval and retries. A badly written one treats the 429 as an error, or worse, retries immediately and makes it worse.

What uses your allowance

More sources than the word "integration" suggests.

• Your own integrations.
• Marketplace apps, some of which are chatty.
• Reporting tools pulling data.
• Migration and bulk-update scripts.

These share one allowance. A migration script running at full speed can starve the app your agents depend on, which is felt immediately in the workspace.

Watch it

Zendesk shows your API usage, and it is worth a look before assuming an integration is broken. A pattern of usage spiking at the same time each day usually points at a scheduled job that could be slowed down for free.

Reducing usage

• Use bulk endpoints rather than one call per record.
• Use incremental exports rather than re-fetching everything.
• Spread scheduled jobs so three do not run at the same minute.
• Cache what does not change often.

Raising it

Possible in some cases, and worth asking about only after you have looked at what is consuming the current allowance. Most accounts that hit the limit have one badly behaved consumer rather than a genuine need for more.

See also

• Keeping an eye on your API usage
• Can I raise my API rate limit
• Can I hit the API limit through normal use of the interface