Skip to main content

Settings

Beginner

Overview

Each canister has a group of possible settings that control its behavior. Only a controller of the canister can read and modify its settings. Currently, there are seven canister setting fields:

  1. controllers: The list of controllers of the canister.
  2. compute_allocation: Amount of compute that this canister has allocated.
  3. memory_allocation: Amount of memory (storage) that this canister has allocated.
  4. freezing_threshold: A safety threshold to protect against canister deletion due to running out of cycles.
  5. reserved_cycles_limit: A safety threshold to protect against spending too many cycles for resource reservation.
  6. wasm_memory_limit: A safety threshold to protect against reaching the 4GiB hard limit of 32-bit Wasm memory.
  7. log_visibility: Controls who can read the canister logs.

Viewing a canister's current settings

There are two ways to read the settings of a canister:

  1. Command line: dfx canister status <canister-id>. Note that this command returns canister settings together with other canister status information.
  2. Programmatically: Call the canister_status endpoint of the IC management canister. The canister settings will be in the settings field of the result.

Note that only a controller of the canister can read canister settings.

An example of using dfx in a command line:

> dfx canister status ajuq4-ruaaa-aaaaa-qaaga-cai
Canister status call result for ajuq4-ruaaa-aaaaa-qaaga-cai.
Controllers: bi7oa-57tth-fsnsf-6xrdb-rruo4-fsiws-xkedg-2uzrc-t6jas-7fo3c-jqe bnz7o-iuaaa-aaaaa-qaaaa-cai
Memory allocation: 0
Compute allocation: 0
Freezing threshold: 2_592_000
Reserved cycles limit: 5_000_000_000_000 Cycles
Wasm memory limit: 0 Bytes
...

An example of programmatically reading canister settings:

actor {
let IC =
actor "aaaaa-aa" : actor {

canister_status : { canister_id : Principal } ->
async ();

};
};

Modifying settings

There are two ways to modify the settings of a canister:

  1. Command line: dfx canister update-settings <canister-id> --<field-name> <field-value>.
  2. Programmatically: Call the update_settings endpoint of the IC management canister.

Note that only a controller of the canister can modify canister settings.

An example of using dfx in a command line:

> dfx canister update-settings ajuq4-ruaaa-aaaaa-qaaga-cai --wasm-memory-limit 3000000000
> dfx canister status ajuq4-ruaaa-aaaaa-qaaga-cai
...
Wasm memory limit: 3_000_000_000 Bytes
...

An example of programmatically modifying canister settings:

actor {
let IC =
actor "aaaaa-aa" : actor {
update_settings : { canister_id : Principal } ->
async ();
};
};

Controllers

The controllers field in a canister's settings contains a list of principals that control the canister. Learn more about canister control.

Compute allocation

By default, canisters are scheduled for execution in a "best-effort" manner. Canisters that require guaranteed execution can get a share of compute capacity by setting compute_allocation in their canister settings.

Compute allocation is expressed in percents and denotes the percentage of an execution core reserved for the canister. For example, compute allocation of 50% means that the canister will get 50% of an execution core. It will be scheduled at least every other round. Compute allocation of 100% means that the canister will be scheduled every round.

Canisters pay a rental fee for compute allocation, meaning that the payment depends on time and the amount of compute allocation regardless whether the canister actually executed or not. For example, an idle canister that has compute allocation of 50% for one hour will be charged base_fee * 50 * 3600, where base_fee is the fee for one percent per second.

The rental fee increases the amount of cycles a canister consumes while idle or frozen. To successfully increase the compute allocation, a canister must have sufficient cycles to run for at least freezing threshold seconds.

The default value of compute allocation is 0%.

Memory allocation

By default, canisters get new memory in a "best-effort" manner. For example, when a canister grows its Wasm or stable memory, the system allocates that memory on demand. That operation can fail if the subnet running the canister is at capacity.

Canisters can pre-allocate memory by setting the memory_allocation field in their canister settings. This field is expressed in bytes and denotes the total amount of memory that the canister pre-allocated. When a canister with a memory allocation configured grows its Wasm or stable memory, the new memory comes from the pre-allocated memory.

Currently, memory_allocation also denotes the maximum amount of memory that the canister is allowed to use in total. For example, if a canister has a memory allocation of 8GiB, then it pre-allocates 8GiBs and cannot exceed that amount. This constraint may be lifted in the future.

Canisters pay a rental fee for memory allocation, meaning that the payment depends on the time duration and the amount of memory allocation, regardless whether the canister actually uses that memory or not. Learn more about payment for memory.

The rental fee increases the amount of cycles a canister consumes while idle or frozen. To successfully increase the memory allocation, a canister must have sufficient cycles to run for at least freezing threshold seconds.

The default value of memory allocation is 0 bytes, which means that there is no default memory allocation.

Freezing threshold

If a canister runs out of cycles, then it gets uninstalled, meaning that its code and data are deleted. A canister will only retain its metadata such as the canister ID, controllers, settings, etc.

The freezing threshold protects against the canister unexpectedly running out of cycles. This field denotes time duration in seconds indicating the minimum duration that the canister should exist without running out of cycles.

When the cycles balance of the canister becomes too low, then the system freezes the canister in order to guarantee its freezing threshold. A frozen canister does not execute messages and pays only for renting resources such as compute allocation, memory allocation, and memory usage.

See also the idle_cycles_burned_per_day field in the canister's status to learn how many cycles a canister consumes in its idle or frozen state.

The default value of the freezing threshold is 2592000 (approximately 30 days).

Reserved cycles limit

In addition to the main cycles balance, a canister has a secondary cycles balance called reserved cycles. The reserved cycles are set aside for future resource payments by the resource reservation mechanism. Learn more about resource reservation.

The reserved_cycles_limit field in a canister's settings serves as the upper limit on reserved cycles. It protects against spending too many cycles in the resource reservation.

The default value is 5 trillion cycles.

Wasm memory limit

A canister has a Wasm memory and a stable memory. Currently, the Wasm memory is 32-bit, which means that it cannot grow beyond the hard limit of 4GiB.

If a canister stores user data in the Wasm memory instead of the stable memory, then its memory usage will increase with the number of users. Using that configuration, a canister could reach 4GiB, at which point it won’t be able to allocate more memory and will stop working. It is also likely that the pre-upgrade hook will fail to allocate, which will make the canister and its data unrecoverable.

Even if the canister doesn’t store user data in Wasm memory, its memory usage may increase due to memory leaks.

The wasm_memory_limit field in a canister's settings provides "best-effort" protection against the canister running out of Wasm memory. The limit is expressed in bytes and it is a soft limit, which means that in some circumstances the canister may grow its Wasm memory above the limit.

Handling of the limit depends on the message type:

  • Update messages: Enforced up to the first await point. If the canister attempts to grow its Wasm memory above the limit, then execution fails with a corresponding error message. After the await point, execution continues in a response callback where the limit is not enforced.
  • Response callback: Not enforced and the canister may grow its Wasm memory above the limit. The motivation for not enforcing the limit is to avoid failing execution after the await point that could lead to subtle bugs in the canister code.
  • Heartbeats and timers: Currently not enforced, but will be enforced in the future after the canister logging feature is available. Without canister logging it is difficult for developers to debug failures in heartbeats and timers.
  • Queries: Not enforced in queries because state changes are not preserved in query execution, so any increase in memory is tentative and will be reverted after the query execution completes.
  • Canister init: Enforced and code installation fails if it grows the Wasm memory above the limit.
  • Canister pre-upgrade: Not enforced in order to allow upgrading the canister to a new version.
  • Canister post-upgrade: Enforced and code upgrades will fail if the new version has Wasm memory usage higher than the limit.

Currently, the default value of the field is 0 bytes, which means that there is no limit. In the future, the default value will be set to 3GiB.

Log visibility

The log_visibility field in a canister's settings can take one of two possible values: controllers or public. Controllers means that only the controllers of the canister can fetch its logs using the fetch_canister_logs endpoint of the management canister. Public means that the logs are public and everyone is able to fetch the logs using the fetch_canister_logs endpoint of the management canister.

The default value of the field is controllers.

Common errors related to canister settings include: