We’re excited to kickstart the Forge journey by extending free use of the Forge platform through 2025. We know the value that apps bring to our customers, and want to support our developer ecosystem in getting started with Forge, whether the app is for your organization, your customers, or for the Atlassian Marketplace.
Forge apps can consume resources up to the quotas or limits outlined below. The Function as a Service (FaaS) and storage quotas scale with your apps install base. The larger the install base (number of seats), the higher the FaaS and storage quotas are. Resource update quotas are applied on a per app basis.
If your app consistently exceeds the quotas or limits, we will contact you. You can also contact us if you think your app needs higher quotas or limits. Learn more about exceeding Forge limits and quotas.
These guidelines apply until the end of December 2024.
In addition to platform quotas and limits, apps may also be affected by product-specific rate limits, such as when making REST API calls to Jira or Confluence.
The FaaS and storage quotas are provided on a per app, per seat basis. For a private app, this is the number of Jira Software and/or Confluence seats that you have licensed. For a free or paid Marketplace app, this is the total number of seats that all your app’s customers have licensed, including evaluations.
As you estimate your app's usage relative to the quotas and limits below, remember that the number of active users at any given point in time will likely be lower than the number of licensed seats for an installation, which would give your app some headroom.
In this example, a paid Marketplace Forge app has 10 500-seat customers, and five 10-seat customers, which is a total of 5,050 seats. The Forge app's weekly quota is shown below.
| FaaS | ||
|---|---|---|
| Invocations (seats) | Runtime (weekly) | Data returned (weekly) |
| 5,050,000 invocations | 20,200 minutes | 99 GB |
| Storage | ||
| Storage capacity (seats) | Reads (weekly) | Writes (weekly) |
| 40,400 MB | 80,800 MB | 20,200 MB |
| Resource updates | ||
| File upload (weekly) | File uploads (weekly) | |
| 150 MB | 500 uploads | |
To estimate how much of the quota the app will use, assume that all of your users are on the standard edition of Jira Software.
The app has two modules that consume FaaS resources: a Jira issue activity panel using custom UI and a product trigger that's listening to the issue created event.
The Jira issue activity panel loads whenever an issue is viewed, and invokes its resolver function once. The resolver takes an average of 500 ms to run, and returns an average of 32 KB to the client. In other words, the app will consume one FaaS invocation, 500 ms of FaaS runtime, and 32 KB of FaaS data returned every time a user views an issue in Jira.
In this example, the view issue event has 70 issue views per seat, per week, for the app’s 10-seat licenses, and 25 issue views per seat for the 500-seat licenses. The app has an estimated 3,500 issue views across the app’s 10-seat licenses and 125,000 issue views across the 500-seat licenses, or 128,500 issue views per week in total.
1 2(70 * 50) + (25 * 5000) = 128,500 weekly invocations (128,500 * 500 ms)/1000/60 = 1,071 minutes (128,500 * 32 KB)/1000/1000 = 3.9 GB
The product trigger fires whenever a Jira issue is created, and takes an average of 2000 ms to run. Note that triggers don't return data, and therefore don't consume the FaaS data quota.
In this example, the create issue event has three issues created per seat for the app’s 10-seat licenses, and one issue created per seat for the 500-seat licenses. The app has an estimated 150 issues created across the app’s 10-seat licenses and 5,000 issues created across the 500-seat licenses, or 5,150 issues created per week total.
1 2(3 * 50) + (1 * 5000) = 5,150 weekly issues (5,150 * 2000 ms)/1000/60 = 172 minutes
Add the totals from the Jira issue panel and product trigger together to get the total FaaS usage of for the app, and calculate the percentage of the quota that you’ll likely use:
1 2128,500 + 5,150 = 133,650 invocations (2.6% of our quota of 5,050,000) 1,071 + 172 = 1,243 minutes (6.2% of our quota of 20,200) 4 GB data returned (4.1% of our quota of 99 GB)
The Jira issue activity panel app uses custom UI, and bundles some static resources that are served to users when they use the app. Note that only resources used in custom UI have quotas applied. Forge JavaScript functions and dependencies are not counted towards the app’s quota. Only deployments to the production environment count towards the quota. Deployments to the development and staging environments do not count towards the quota.
In this example, the static resources include an HTML file (1 MB), a JavaScript file (1 MB), a CSS file (1 MB), a PNG file (1 MB), and a GIF (1 MB); which is a total of 5 files and 5 MB.
Each time the app is deployed, the following quota is consumed:
1 25 * 1 MB = 5 MB uploaded (3% of the 150 MB quota) 5 files uploaded (1% of the 500 files quota)
This app consumes the weekly “File uploads (MB)” resource quota at a greater rate than our “File uploads (file count)” resource quota. Divide the total quota by the MB consumed per deploy to determine how many times the app can be deployed weekly before exceeding the quota.
1 2150 MB quota per week / 5 MB deployment size = 30 production deployments per week
If the app consumes the “File uploads (file count)” quota faster than the “File uploads (MB)” quota, the number of files being deployed is the limiting factor in how many deployments can be performed in a given week.
The quotas listed below apply to an app installation; E.g. the example app, installed on example.atlassian.net in the production environment.
There are three tiers for quotas, depending on how your app is deployed:
The FaaS quotas are consumed when the function is invoked in any environment, and scale with the number of seats in your app’s install base. Runtime is metered by millisecond or part thereof. Data returned is metered by KB or part thereof. These quotas are refreshed weekly.
| Paid apps | Free apps | Distributed apps | ||||
|---|---|---|---|---|---|---|
| First 100 seats, total | Per seat thereafter | Up to 100 seats, total | Per seat thereafter | Up to 100 seats, total | Per seat thereafter | |
| Invocations (weekly) | 100,000 invocations | 1,000 invocations | 50,000 invocations | 500 invocations | 50,000 invocations | 500 invocations |
| Runtime (weekly) | 400 minutes | 4 minutes | 200 minutes | 2 minutes | 200 minutes | 2 minutes |
| Data returned (weekly) | 2,000 MB | 20 MB | 1,000 MB | 10 MB | 1,000 MB | 10 MB |
These quotas apply to the static resources packaged with your app that are used by custom UI functions. Note that Forge JavaScript functions and dependencies are not counted towards the app’s quota. Resource quotas are consumed per deployment to your production environment; deployments to development and staging environments are unmetered. These quotas are refreshed weekly.
| Paid apps | Free apps | Distributed apps | |
|---|---|---|---|
| Per app | Per app | Per app | |
| File capacity (weekly) | 150 MB | 75 MB | 75 MB |
| Files uploaded (weekly) | 500 files | 250 files | 250 files |
These quotas apply to the use of Forge hosted storage in any environment, and scale with the number of seats in your app's install base.
If an installation decreases its seat count, quotas will currently not decrease accordingly. This may change in the future, at which point we will provide guidance on how to react to these events. Meanwhile, follow your GDPR responsibilities with respect to storing, reporting and erasing user data.
Currently, the Storage API is not intended for large objects such as media files.
For the time being, consider storing large objects as product attachments (see the issue attachment API for Jira Cloud or the content - attachment API for Confluence Cloud), or using an external storage solution such as Amazon S3.
As with all quotas and limits, if you suspect the Storage API quotas will prevent you from implementing your Forge app, contact us to discuss your use case.
Storage capacity (MB) is the total amount of storage provisioned per seat for each installation of your app. These quotas are tracked in terms of raw format size.
The Storage API provides a way to store sensitive credentials (secrets), which is subject to a separate total storage quota from general storage.
| Paid apps | Free apps | Distributed apps | ||||
|---|---|---|---|---|---|---|
| First 100 seats, total | Per seat thereafter | Up to 100 seats, total | Per seat thereafter | Up to 100 seats, total | Per seat thereafter | |
| Storage capacity | 1200 MB | 12 MB | 600 MB | 6 MB | 600 MB | 6 MB |
| Secret storage capacity | 200 MB | 2 MB | 100 MB | 1 MB | 100 MB | 1 MB |
Read and write capacity quotas are the amount of data that can be transferred in or out of the Storage API each week, per seat. These quotas are refreshed weekly, and are tracked in terms of raw format size.
| Paid apps | Free apps | Distributed apps | ||||
|---|---|---|---|---|---|---|
| First 100 seats, total | Per seat thereafter | Up to 100 seats, total | Per seat thereafter | Up to 100 seats, total | Per seat thereafter | |
| Read capacity per week | 2400 MB | 24 MB | 1200 MB | 12 MB | 1200 MB | 12 MB |
| Write capacity per week | 600 MB | 6 MB | 300 MB | 3 MB | 300 MB | 3 MB |
The Forge platform has additional limits that cannot be scaled with your app's usage. To prevent your app from exceeding these limits, you may need to tune your app or remove some resources. Learn more about exceeding Forge limits and quotas.
An app can be invoked by users, webtriggers, or scheduled triggers. The following limits apply to all three:
| Resource | Limit | Description |
|---|---|---|
| Invocation rate limit | 500 | Maximum number of invocations per 25-second sliding window. That is, an app reaches this limit when it is invoked 500 times within the last 25 seconds. |
| Runtime seconds (also includes UI modules invoked by Forge remote) | 25 | Maximum runtime permitted before the app is stopped. |
| Runtime seconds (async events) | 900 | This applies to function modules that are only referenced by consumer modules. Default timeout is 55 seconds. Use timeoutSeconds to extend it. |
| Runtime seconds (events invoked by Forge remote) | 5 | Maximum runtime permitted before the app is stopped. This applies to remote back ends receiving events from the Atlassian platform. |
| Log lines per invocation | 100 per runtime minute (rounded up) |
Maximum number of log entries for an invocation.
The limit is calculated based on the function timeout, specified by timeoutSeconds, rounded up per minute.
|
| Log size per invocation | 200 KB | Maximum size of all log line data generated per invocation. |
| Log file size per download | 100 MB | Maximum file size of filtered logs per download. |
| Log lines per download | 96,000 | Maximum number of log entries per download. |
| Network requests | 100 per runtime minute (rounded up) |
Number of network requests per invocation, excluding those made using requestJira or requestConfluence.
The limit is calculated based on the function timeout, specified by timeoutSeconds, rounded up per minute.
|
| Memory | 512MB | Available memory per invocation. |
If your app is still running on the previous runtime version, it only has 128MB of available memory per invocation.
We strongly recommend all Forge developers to use the current runtime. For details about the previous runtime (including instructions for migration), see Upgrading from legacy runtime.
The limits listed below apply to resources bundled with an app.
| Category | Resource | Limit | Description |
|---|---|---|---|
| Custom UI | Bundle files | 5000 | Maximum number of files declared in a single resource bundle. |
| Bundle size | 50 | Maximum bundle size in megabytes (MB) for a resource. | |
| UI Kit | Bundle files | 5000 | Maximum number of files declared in a single bundle. |
| Bundle size | 50 | Maximum bundle size in megabytes (MB) for a resource. |
When using the app storage API, we strongly recommend that you:
The Atlassian Cloud backs up the entire hosted storage for disaster recovery. This includes content stored from the Forge storage API.
If an installation of your app exceeds these limits due to bulk processing (for example, triggered by a bulk issue update in Jira), consider using the Async events API to queue your app's interactions with the Storage API. See Queue app interactions with storage API for guidance.
The limits listed below apply to the Key-Value Store or Custom Entity Store for each installation of your app.
| Resource | Limit | Description |
|---|---|---|
| Delete operations | 400 | Maximum number of delete operations per 20 seconds |
| Query operations | 200 | Maximum number of query operations per 20 seconds |
| Read operations | 1000 | Maximum number of read operations per 20 seconds |
| Update operations | 1000 | Maximum number of update operations per 20 seconds |
Forge hosted storage has the following limits on keys and values.
You don't need to include any identifiers for apps or installations in your key.
Internally, Forge automatically prepends an identifier to every key, mapping it to the right app and installation. This lets you use the full key length without risking conflicts across apps or installations.
| Resource | Limit | Description |
|---|---|---|
| Key length | 500 | Maximum length of a key |
| Key format | /^(?!\s+$)[a-zA-Z0-9:._\s-#]+$/ | Keys must match the regex |
| Value size | 240 KiB | Maximum size of a single persisted value (in RAW) |
| Object depth | 31 | Maximum object depth permitted |
Custom entities (used for complex queries) are subject to the following limits:
| Category | Requirements | Limits |
|---|---|---|
| Entity | Entity names:
In addition, an app must not have duplicate entity names. |
|
| Attribute |
Attribute names must follow the regex [_A-Za-z][_0-9A-Za-z]*.
| Attribute names have a maximum length of 64 characters. |
| Index |
Index names must contain only the following characters
In addition, each index name within an entity must be unique.
|
|
| Keys | A key should:
| A key can contain a maximum of 500 characters. |
In addition, each complex query
can only have a maximum of 100 conditions
(for example, beginsWith, contains, and isGreaterThan). This limit applies to conditions used in
where,
andFilter, and
orFilter filters.
Forge Cache is now available as part of Forge Early Access Program (EAP). To start testing this feature, sign up using this form.
Forge Cache is an experimental feature offered to selected users for testing and feedback purposes. This feature is unsupported and subject to change without notice. Do not use Forge Cache in apps that handle sensitive information and customer data.
For more details, see Forge EAP, Preview, and GA.
Each installation of your app is subject to the limits of Cache API.
| Resource | Limit | Description |
|---|---|---|
| Key length | 500 characters | The maximum length of a key. |
| Key format | Regex: /^(?!\s+$)[a-zA-Z0-9:._\s-#]+$/ | Keys must match the regex. |
| Value size | 240 KiB | The maximum size of a single stored value. |
| Time-to-live (TTL) | Between 30s and 3600s | The duration (in seconds) that the key will exist in the cache before it's automatically deleted. |
The Cache API will only work on functions where @forge/api is also supported. Otherwise, you will need to invoke your cache function through a resolver (like all back end functions).
| Limit | Value | Description |
|---|---|---|
| Write operations | 6,000 | Maximum number of writes that an app can make to Forge Cache per minute. |
| Read operations | 12,000 | Maximum number of reads that an app can make to Forge Cache per minute. |
The Preview version of Forge SQL currently does not support data residency. By using Forge SQL, you acknowledge that Section 6 (Data Residency) of the Forge Terms does not apply to data stored using Forge SQL, and you must not use Forge SQL for any applications that require data residency. For more details, please see the Data residency limitations section.
Forge SQL offers a multi-tenant SQL solution offering tenant isolation and stable query performance. To do so, we add a few constraints over and above TiDB’s limitations:
Foreign keys are not supported. You can still perform JOIN operations, but DELETE operations will not be cascaded.
Each READ operation has a 5-second timeout configured.
Forge SQL API (for non-READ statements) will timeout after 10 seconds of execution.
Each QUERY operation has a memory quota set to 16 MB.
Within each minute, all concurrent invocations share a total query execution time of 62.5 seconds (per installation).
Each SQL statement can only contain a single query.
The following limits also apply:
| Resource | Limit | Description |
|---|---|---|
| Input payload size | 256 KiB | Maximum size of a single request from a Forge app to execute a single SQL query. Forge will return an error if this is exceeded. |
| Response payload size | 5 MiB | Maximum size of a response to any SQL query executed by a Forge app. Forge will return an error if a response exceeds this limit. |
| Number of DDL operations | 25 | Maximum number of Data Definition Language (DDL) operations each installation can perform per minute. See Manage your database schema for more details. |
| Number of DML operations | 3000 | Maximum number of Data Modification Language (DML) operations each installation can perform per minute. See Execute SQL operations for more details. |
Forge SQL currently doesn’t support data residency. We plan to add data residency support in a future Forge SQL update.
For now, you understand and acknowledge that you are instructing Atlassian to provision all customer databases for Forge SQL in the US-WEST region.
Once Forge SQL adds data residency features, we will migrate all Production databases provisioned by SQL to their respective products' regions. The migration will require downtime, so we will announce this at least 30 days in advance (allowing affected developers and customers to plan accordingly).
Forge SQL has not yet undergone external assessment for SOC 2 or ISO certification. As we continue development on Forge SQL, we will aim to include it in our standard audit certification reporting cycle.
The Runs on Atlassian program requires that you only use storage with data residency features. As such, apps using Forge SQL are ineligible for this program for now.
Such apps will become eligible once Forge SQL adds data residency features.
If you add Forge SQL to an existing app, admins of that app’s current installations must review and consent before updating.
As such, adding Forge SQL to an existing app will require a major version upgrade. This will be triggered through the sql module (which is required to enable Forge SQL on an app).
You must inform your app’s users (for example, through release notes) about the impact of adopting Forge SQL, particularly in relation to data residency.
The limits listed below apply to the Async events API for each installation of your app.
| Resource | Limit | Description |
|---|---|---|
| Event per request | 50 | Maximum number of events pushed in a single request. |
| Event per minute | 500 | Maximum number of events pushed in one minute. |
| Payload size | 200 KB | Maximum combined payload size of events in single request. |
| Payload size for long running functions | 100 KB | Maximum size of an individual event. This limit only applies to functions specifying a timeout greater than 55 seconds. |
| Retry data size for long running functions | 2 KB | Maximum size of This limit only applies to functions specifying a timeout greater than 55 seconds. |
| Cyclic invocation limit | 1000 | An event resolver can push more events to the queue thus creating a cycle. This limit applies to the maximum cyclic depth up to which events resolver can push more events to the queue. |
The limits listed below apply to an app.
| Resource | Limit | Description |
|---|---|---|
| App description | 1,000 | Maximum number of characters that the app description can have. |
| App name | 1 to 60 | The app name must be between the character limits. |
| App size | 100 | Maximum app size in megabytes (MB). |
| Base URL | 2,048 | Maximum number of characters that the app baseUrl can have. |
| Modules per app | 100 | Maximum number of unique modules that can be declared in a single app manifest. |
| Resources per app | 10 | Maximum number of unique resources that can be declared in a single app manifest. |
| Environments per app | 25 | Maximum number of environments in a single app. |
| Alerts per app | 5 | Maximum number of alerts that can be created for a single app. |
The limits listed below apply to a developer's account.
| Resource | Limit | Description |
|---|---|---|
| Apps | 100 | Maximum number of apps that can be created or owned by a single developer. |
If your app exceeds any quotas or limits, you can often fix the issue yourself:
Most app limits are enforced through Forge CLI validation, where you'll immediately receive an error. Most errors have trivial fixes, such as shortening your app's name.
For resource-bound limit errors (such as the total number of apps), you need to
remove the relevant resources. You can uninstall an app with the forge uninstall command.
Once you remove all installations of an app, delete it from the
developer console.
We understand that some quotas can be hard to monitor; as such, we are working on better monitoring tools for future releases. Meanwhile, if we detect that your app consistently exceeds our quotas or limits, we'll first contact you to understand why. If your app needs higher storage capacity or a higher cyclic invocation limit, you can contact us through Developer and Marketplace support. If your app needs an increase in other quotas or limits, please let us know through the Forge Jira Project.
We aim to unblock reasonable use cases, and we will work with you to achieve this. However, repeated or prolonged failure to address requests to comply with quotas may result in further action being taken (such as app suspension).
An app may be temporarily suspended if it negatively impacts the Forge platform, regardless of whether it’s in breach of any quotas or limits.
Suspended apps cannot be:
Invoked: Suspended apps cannot be invoked by any existing installations. Invoking
a suspended app will return an App is currently unavailable, please try again later error.
Installed: Attempting to install a suspended app will return an
App installation is not available while the app is suspended error.
Deployed: Running forge deploy command on a suspended app will return an
App management is unavailable while the app is suspended error.
If your app is suspended, we'll submit a ticket through Developer and Marketplace support and mention you, so that you can help us fix the issue as soon as possible.
If you're not a Jira user, we’ll email you a link to the issue.
If have an issue with quotas or limits but haven't been contacted by our team, you can seek assistance from the Forge developer community.
See the Forge Terms for more information.
Rate this page: