Skip to content

feat: Support for canary deployments in gRPC controller #1081

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
tylbees wants to merge 1 commit into
base: main
Choose a base branch
from
Open

feat: Support for canary deployments in gRPC controller #1081

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
52 changes: 51 additions & 1 deletion projects/isolate_proto/src/isolate_proto/controller.proto
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,6 +62,14 @@ service IsolateController {
// Build (or warm the cache for) an environment without running anything in it.
// Streams build logs and terminal status; reuses the same caches as Run/RegisterApplication.
rpc BuildEnvironment (BuildEnvironmentRequest) returns (stream BuildEnvironmentResult) {}
// Update application canary config
rpc UpdateCanary(UpdateCanaryRequest) returns (UpdateCanaryResponse) {}
// Rollback (cancel) application canary
rpc RollbackCanary(RollbackCanaryRequest) returns (RollbackCanaryResponse) {}
// Commit (roll forward) application canary
rpc CommitCanary(CommitCanaryRequest) returns (CommitCanaryResponse) {}
// Get application canary info
rpc CanaryInfo(CanaryInfoRequest) returns (CanaryInfoResponse) {}
}

message HostedMap {
Expand Down Expand Up @@ -290,6 +298,7 @@ enum ApplicationAuthMode {
enum DeploymentStrategy {
RECREATE = 0;
ROLLING = 1;
CANARY = 2;
Comment on lines 298 to +301

@chatgpt-codex-connector chatgpt-codex-connector Bot Jun 11, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Add CANARY to SDK strategy handling

Adding CANARY to the shared DeploymentStrategy enum makes it a valid value for fields like RegisterApplicationRequest.deployment_strategy and SetAliasRequest.deployment_strategy, but the Python SDK/CLI still only allows "recreate" and "rolling" (projects/fal/src/fal/sdk.py:52, projects/fal/src/fal/sdk.py:462-484, and CLI choices in projects/fal/src/fal/cli/apps.py:385-386). In those paths, a user trying to select the newly advertised canary strategy is rejected by argparse or hits DeploymentStrategy[deployment_strategy.upper()] with no enum member, so the public Python tooling cannot send this new proto value.

Useful? React with 👍 / 👎.

}

message ApplicationHealthCheckConfig {
Expand Down Expand Up @@ -511,6 +520,7 @@ message AliasInfo {
optional int32 scaling_delay_seconds = 15;
optional string environment_name = 16;
bool is_rolling = 17;
optional CanaryInfo canary = 18;
}

message SetSecretRequest {
Expand Down Expand Up @@ -595,7 +605,6 @@ message StopRunnerResponse {
// Empty. For future use.
}


message KillRunnerRequest {
string runner_id = 1;
}
Expand Down Expand Up @@ -663,3 +672,44 @@ message DeleteEnvironmentRequest {
message DeleteEnvironmentResponse {
// Empty. For future use.
}

message CanaryInfo {
optional bool active = 1;
optional string revision = 2;
optional uint32 ratio = 3; // 0-100
optional uint32 runners = 4;
}

message UpdateCanaryRequest {
string alias = 1;
uint32 ratio = 2; // 0-100
}

message UpdateCanaryResponse {
CanaryInfo info = 1;
}

message RollbackCanaryRequest {
string alias = 1;
}

message RollbackCanaryResponse {
AliasInfo alias = 1;
}

message CommitCanaryRequest {
string alias = 1;
DeploymentStrategy strategy = 2;
}

message CommitCanaryResponse {
AliasInfo alias = 1;
}

message CanaryInfoRequest {
string alias = 1;
}

message CanaryInfoResponse {
CanaryInfo info = 1;
Comment on lines +713 to +714

@chatgpt-codex-connector chatgpt-codex-connector Bot Jun 11, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Badge Regenerate the canary response schema consistently

When clients generate code from this committed .proto, CanaryInfoResponse is a wrapper containing info, but the committed Python artifacts describe CanaryInfoResponse itself as the scalar canary-info message with active/revision/ratio/runners fields (for example controller_pb2.pyi defines those fields directly). In a mixed environment where a non-Python server or client is generated from controller.proto, the CanaryInfo RPC response wire shape disagrees with the published Python package, so Python clients will ignore the length-delimited info field or other clients will miss the direct scalar fields. Please regenerate the artifacts from this schema or rename/split the messages so the source proto and generated files match.

Useful? React with 👍 / 👎.

}
140 changes: 78 additions & 62 deletions projects/isolate_proto/src/isolate_proto/controller_pb2.py
View file
Open in desktop

Large diffs are not rendered by default.

164 changes: 162 additions & 2 deletions projects/isolate_proto/src/isolate_proto/controller_pb2.pyi
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,11 +47,13 @@ class _DeploymentStrategyEnumTypeWrapper(google.protobuf.internal.enum_type_wrap
DESCRIPTOR: google.protobuf.descriptor.EnumDescriptor
RECREATE: _DeploymentStrategy.ValueType # 0
ROLLING: _DeploymentStrategy.ValueType # 1
CANARY: _DeploymentStrategy.ValueType # 2

class DeploymentStrategy(_DeploymentStrategy, metaclass=_DeploymentStrategyEnumTypeWrapper): ...

RECREATE: DeploymentStrategy.ValueType # 0
ROLLING: DeploymentStrategy.ValueType # 1
CANARY: DeploymentStrategy.ValueType # 2
global___DeploymentStrategy = DeploymentStrategy

class _RetryCondition:
Expand Down Expand Up @@ -1429,6 +1431,7 @@ class AliasInfo(google.protobuf.message.Message):
SCALING_DELAY_SECONDS_FIELD_NUMBER: builtins.int
ENVIRONMENT_NAME_FIELD_NUMBER: builtins.int
IS_ROLLING_FIELD_NUMBER: builtins.int
CANARY_FIELD_NUMBER: builtins.int
alias: builtins.str
revision: builtins.str
auth_mode: global___ApplicationAuthMode.ValueType
Expand All @@ -1448,6 +1451,8 @@ class AliasInfo(google.protobuf.message.Message):
scaling_delay_seconds: builtins.int
environment_name: builtins.str
is_rolling: builtins.bool
@property
def canary(self) -> global___CanaryInfoResponse: ...
def __init__(
self,
*,
Expand All @@ -1468,9 +1473,12 @@ class AliasInfo(google.protobuf.message.Message):
scaling_delay_seconds: builtins.int | None = ...,
environment_name: builtins.str | None = ...,
is_rolling: builtins.bool = ...,
canary: global___CanaryInfoResponse | None = ...,
) -> None: ...
def HasField(self, field_name: typing_extensions.Literal["_concurrency_buffer", b"_concurrency_buffer", "_concurrency_buffer_perc", b"_concurrency_buffer_perc", "_environment_name", b"_environment_name", "_request_timeout", b"_request_timeout", "_scaling_delay_seconds", b"_scaling_delay_seconds", "_startup_timeout", b"_startup_timeout", "concurrency_buffer", b"concurrency_buffer", "concurrency_buffer_perc", b"concurrency_buffer_perc", "environment_name", b"environment_name", "request_timeout", b"request_timeout", "scaling_delay_seconds", b"scaling_delay_seconds", "startup_timeout", b"startup_timeout"]) -> builtins.bool: ...
def ClearField(self, field_name: typing_extensions.Literal["_concurrency_buffer", b"_concurrency_buffer", "_concurrency_buffer_perc", b"_concurrency_buffer_perc", "_environment_name", b"_environment_name", "_request_timeout", b"_request_timeout", "_scaling_delay_seconds", b"_scaling_delay_seconds", "_startup_timeout", b"_startup_timeout", "active_runners", b"active_runners", "alias", b"alias", "auth_mode", b"auth_mode", "concurrency_buffer", b"concurrency_buffer", "concurrency_buffer_perc", b"concurrency_buffer_perc", "environment_name", b"environment_name", "is_rolling", b"is_rolling", "keep_alive", b"keep_alive", "machine_types", b"machine_types", "max_concurrency", b"max_concurrency", "max_multiplexing", b"max_multiplexing", "min_concurrency", b"min_concurrency", "request_timeout", b"request_timeout", "revision", b"revision", "scaling_delay_seconds", b"scaling_delay_seconds", "startup_timeout", b"startup_timeout", "valid_regions", b"valid_regions"]) -> None: ...
def HasField(self, field_name: typing_extensions.Literal["_canary", b"_canary", "_concurrency_buffer", b"_concurrency_buffer", "_concurrency_buffer_perc", b"_concurrency_buffer_perc", "_environment_name", b"_environment_name", "_request_timeout", b"_request_timeout", "_scaling_delay_seconds", b"_scaling_delay_seconds", "_startup_timeout", b"_startup_timeout", "canary", b"canary", "concurrency_buffer", b"concurrency_buffer", "concurrency_buffer_perc", b"concurrency_buffer_perc", "environment_name", b"environment_name", "request_timeout", b"request_timeout", "scaling_delay_seconds", b"scaling_delay_seconds", "startup_timeout", b"startup_timeout"]) -> builtins.bool: ...
def ClearField(self, field_name: typing_extensions.Literal["_canary", b"_canary", "_concurrency_buffer", b"_concurrency_buffer", "_concurrency_buffer_perc", b"_concurrency_buffer_perc", "_environment_name", b"_environment_name", "_request_timeout", b"_request_timeout", "_scaling_delay_seconds", b"_scaling_delay_seconds", "_startup_timeout", b"_startup_timeout", "active_runners", b"active_runners", "alias", b"alias", "auth_mode", b"auth_mode", "canary", b"canary", "concurrency_buffer", b"concurrency_buffer", "concurrency_buffer_perc", b"concurrency_buffer_perc", "environment_name", b"environment_name", "is_rolling", b"is_rolling", "keep_alive", b"keep_alive", "machine_types", b"machine_types", "max_concurrency", b"max_concurrency", "max_multiplexing", b"max_multiplexing", "min_concurrency", b"min_concurrency", "request_timeout", b"request_timeout", "revision", b"revision", "scaling_delay_seconds", b"scaling_delay_seconds", "startup_timeout", b"startup_timeout", "valid_regions", b"valid_regions"]) -> None: ...
@typing.overload
def WhichOneof(self, oneof_group: typing_extensions.Literal["_canary", b"_canary"]) -> typing_extensions.Literal["canary"] | None: ...
@typing.overload
def WhichOneof(self, oneof_group: typing_extensions.Literal["_concurrency_buffer", b"_concurrency_buffer"]) -> typing_extensions.Literal["concurrency_buffer"] | None: ...
@typing.overload
Expand Down Expand Up @@ -2018,3 +2026,155 @@ class DeleteEnvironmentResponse(google.protobuf.message.Message):
) -> None: ...

global___DeleteEnvironmentResponse = DeleteEnvironmentResponse

@typing_extensions.final
class UpdateCanaryRequest(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

ALIAS_FIELD_NUMBER: builtins.int
RATIO_FIELD_NUMBER: builtins.int
alias: builtins.str
ratio: builtins.int
"""0-100"""
def __init__(
self,
*,
alias: builtins.str = ...,
ratio: builtins.int = ...,
) -> None: ...
def ClearField(self, field_name: typing_extensions.Literal["alias", b"alias", "ratio", b"ratio"]) -> None: ...

global___UpdateCanaryRequest = UpdateCanaryRequest

@typing_extensions.final
class UpdateCanaryResponse(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

INFO_FIELD_NUMBER: builtins.int
@property
def info(self) -> global___CanaryInfoResponse: ...
def __init__(
self,
*,
info: global___CanaryInfoResponse | None = ...,
) -> None: ...
def HasField(self, field_name: typing_extensions.Literal["info", b"info"]) -> builtins.bool: ...
def ClearField(self, field_name: typing_extensions.Literal["info", b"info"]) -> None: ...

global___UpdateCanaryResponse = UpdateCanaryResponse

@typing_extensions.final
class RollbackCanaryRequest(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

ALIAS_FIELD_NUMBER: builtins.int
alias: builtins.str
def __init__(
self,
*,
alias: builtins.str = ...,
) -> None: ...
def ClearField(self, field_name: typing_extensions.Literal["alias", b"alias"]) -> None: ...

global___RollbackCanaryRequest = RollbackCanaryRequest

@typing_extensions.final
class RollbackCanaryResponse(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

ALIAS_FIELD_NUMBER: builtins.int
@property
def alias(self) -> global___AliasInfo: ...
def __init__(
self,
*,
alias: global___AliasInfo | None = ...,
) -> None: ...
def HasField(self, field_name: typing_extensions.Literal["alias", b"alias"]) -> builtins.bool: ...
def ClearField(self, field_name: typing_extensions.Literal["alias", b"alias"]) -> None: ...

global___RollbackCanaryResponse = RollbackCanaryResponse

@typing_extensions.final
class CommitCanaryRequest(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

ALIAS_FIELD_NUMBER: builtins.int
STRATEGY_FIELD_NUMBER: builtins.int
alias: builtins.str
strategy: global___DeploymentStrategy.ValueType
def __init__(
self,
*,
alias: builtins.str = ...,
strategy: global___DeploymentStrategy.ValueType = ...,
) -> None: ...
def ClearField(self, field_name: typing_extensions.Literal["alias", b"alias", "strategy", b"strategy"]) -> None: ...

global___CommitCanaryRequest = CommitCanaryRequest

@typing_extensions.final
class CommitCanaryResponse(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

ALIAS_FIELD_NUMBER: builtins.int
@property
def alias(self) -> global___AliasInfo: ...
def __init__(
self,
*,
alias: global___AliasInfo | None = ...,
) -> None: ...
def HasField(self, field_name: typing_extensions.Literal["alias", b"alias"]) -> builtins.bool: ...
def ClearField(self, field_name: typing_extensions.Literal["alias", b"alias"]) -> None: ...

global___CommitCanaryResponse = CommitCanaryResponse

@typing_extensions.final
class CanaryInfoRequest(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

ALIAS_FIELD_NUMBER: builtins.int
alias: builtins.str
def __init__(
self,
*,
alias: builtins.str = ...,
) -> None: ...
def ClearField(self, field_name: typing_extensions.Literal["alias", b"alias"]) -> None: ...

global___CanaryInfoRequest = CanaryInfoRequest

@typing_extensions.final
class CanaryInfoResponse(google.protobuf.message.Message):
DESCRIPTOR: google.protobuf.descriptor.Descriptor

ACTIVE_FIELD_NUMBER: builtins.int
REVISION_FIELD_NUMBER: builtins.int
RATIO_FIELD_NUMBER: builtins.int
RUNNERS_FIELD_NUMBER: builtins.int
active: builtins.bool
revision: builtins.str
ratio: builtins.int
"""0-100"""
runners: builtins.int
def __init__(
self,
*,
active: builtins.bool | None = ...,
revision: builtins.str | None = ...,
ratio: builtins.int | None = ...,
runners: builtins.int | None = ...,
) -> None: ...
def HasField(self, field_name: typing_extensions.Literal["_active", b"_active", "_ratio", b"_ratio", "_revision", b"_revision", "_runners", b"_runners", "active", b"active", "ratio", b"ratio", "revision", b"revision", "runners", b"runners"]) -> builtins.bool: ...
def ClearField(self, field_name: typing_extensions.Literal["_active", b"_active", "_ratio", b"_ratio", "_revision", b"_revision", "_runners", b"_runners", "active", b"active", "ratio", b"ratio", "revision", b"revision", "runners", b"runners"]) -> None: ...
@typing.overload
def WhichOneof(self, oneof_group: typing_extensions.Literal["_active", b"_active"]) -> typing_extensions.Literal["active"] | None: ...
@typing.overload
def WhichOneof(self, oneof_group: typing_extensions.Literal["_ratio", b"_ratio"]) -> typing_extensions.Literal["ratio"] | None: ...
@typing.overload
def WhichOneof(self, oneof_group: typing_extensions.Literal["_revision", b"_revision"]) -> typing_extensions.Literal["revision"] | None: ...
@typing.overload
def WhichOneof(self, oneof_group: typing_extensions.Literal["_runners", b"_runners"]) -> typing_extensions.Literal["runners"] | None: ...

global___CanaryInfoResponse = CanaryInfoResponse
Loading