What Atlassian changed
On 13 November 2026 Atlassian posted a notice in its Jira Cloud platform changelog. More Jira Cloud REST API operations will now enforce the configuration limits that Jira already applies in its admin screens. If a request would take a site past one of those limits, the API returns HTTP 422 Unprocessable Entity. The rollout was planned to start on 11 or 12 November. Enforcement is switched on one site at a time, so some tenants will get the new behaviour before others. Atlassian says plainly that some requests which used to succeed will now be rejected.
The changelog lists the affected operations:
- adding options to a custom field context
- creating issue security schemes with their initial levels
- creating permission schemes with their initial grants
- creating priorities
- editing an issue in a way that adds new components to a space
- creating a project from a template when the template's contents would go over a limit
Assigning components that already exist in a space isn't affected.
The limits behind the errors
The numbers come from Atlassian's data limits and guardrails page. Since March 2026, Jira has capped each space at 700 fields and 150 work types. A second set of limits began rolling out gradually from September 2026:
- 20,000 options per field
- 50 work item security levels per space
- 50 grants per permission
- 15,000 releases per space
- 150 workflows per workflow scheme
- 200 statuses per workflow
- 10,000 components per space
- 100 priorities per space
Atlassian says the date enforcement starts can vary by entity and by site.
Atlassian first announced the September limits in a March 2026 Atlassian Community post. The post linked them to keeping large sites fast, because very large Jira sites become harder to optimise. Until now, most people only met the limits as an in-product message when an administrator tried to add one item too many. The 13 November change applies the same rule to scripts, integrations and apps that write configuration through the API.
Jira handles a breached limit cautiously. The guardrails page says existing configuration keeps working and nothing is deleted or archived. End users can still create and update work items. If a bulk action would breach a limit, Jira rejects the whole batch, so none of it is saved.
Why service desks should care
Many NZ organisations on Jira Service Management have built background automation around it over the years. Typical examples include:
- a nightly job that copies customer or site names from an ERP or CRM into a select-list custom field
- an onboarding script that creates a project from a template for each new client
- an asset integration that adds a component for every application it finds
The 422 enforcement now covers all of these.
The failure mode is what concerns us. A sync job that has added field options every night for five years can reach 20,000 options without anyone noticing, because agents can still use the field. Once the API enforces the limit, the job's next write fails. If the integration logs the error and carries on, new customers quietly stop appearing in the request form. The first anyone hears of it may be an agent who can't find the right value. If the integration retries over and over, it adds load and fills the logs with noise. Atlassian's guidance is to treat a 422 as a configuration-limit error, read the response body to see which limit was hit, and not retry the same request automatically.
We'd check permission grants first. Actonic's August 2026 analysis calls the limit of 50 grants per permission the tightest of the set. Actonic also says it's the one most likely to be reached through ordinary admin work. Releases are worth checking too. Actonic's example is a CI pipeline that creates a version for every build and reaches 15,000 in about sixteen months. The 13 November changelog doesn't include the release endpoint among the newly enforced operations, but the guardrails page lists the release limit among those Jira is enforcing from September.
What's still unknown
Atlassian hasn't published a per-site schedule, so from outside there's no way to tell when a given tenant got the change. The changelog doesn't say whether more endpoints will follow. Its wording about extending enforcement to more operations suggests the list could grow. Marketplace apps that create configuration for you hit the same 422 responses when they call the affected endpoints. Ask each vendor whether their app's error handling has been updated.
Practical next steps
- List every script, integration and Marketplace app with admin-level API access to your Jira site. Include low-code tools and the service account tokens behind them, because those often get missed.
- Compare your current counts with the limits. Premium and Enterprise sites can use Site Optimizer. On Standard plans, Actonic notes that the public REST API returns a total for each entity, and the first page of results is enough to get it.
- Search your integration code for the endpoints listed in the changelog notice. Add explicit handling for HTTP 422 that logs the limit named in the response body, alerts a person and stops retrying.
- Clean up before you reach a limit. According to Actonic's summary, Atlassian's tools let administrators bulk-disable unused field options, delete unused security levels, and remove suspended users and unused groups from permission grants.
- Redesign patterns that grow without end. For a select list holding thousands of customer names, an Assets object or a lookup against the source system usually works better.
- Add the limits to your change process, so new projects and templates are checked against them before go-live. Our ITIL 4 framework page explains how we approach change enablement and configuration management. Integration redesign falls under our process and automation work.
Sources
- Jira Cloud Platform changelog (Atlassian Developer)
- Jira Service Management Cloud changelog (Atlassian Developer)
- Data limits and guardrails in Jira Cloud (Atlassian Support)
- Growing our investment in a faster, cleaner Jira (Atlassian Community, reader mirror)
- Jira data limits, September 2026: what they cap and how to check your instance (Actonic)
Want to know what this means for your organisation? Our team in Auckland can show you your top three opportunities in a free 30 minute assessment.
Get my free assessment How we help