mirror of
https://github.com/mudler/LocalAI.git
synced 2026-10-28 17:18:49 -04:00
Ettore Di Giacinto
1604a67c1f
Add the add-on runtime to LocalAI: manifests, the host operation broker, admin gating by account role, the add-on UI bundle builder and the SDK for add-on pages. Move the navigation to Chat, Models, Add-ons and Settings, with pinned add-ons in the sidebar and an Add-ons drawer. Ship the first-party add-ons (voice, face, fine-tuning, quantization) as packages for linux/amd64, linux/arm64 and darwin/arm64. Intel macOS (darwin/amd64) is enabled in the platform files. Squashed from the review branch. The tree is unchanged; the pre-squash history is kept on a local backup branch. Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
50 lines
2.3 KiB
Go
50 lines
2.3 KiB
Go
// SPDX-License-Identifier: MIT
|
|
package contract
|
|
|
|
import (
|
|
"fmt"
|
|
"reflect"
|
|
"time"
|
|
)
|
|
|
|
// Limits are host configuration, not addon-controlled manifest claims. Durations
|
|
// use Go time.Duration (nanoseconds in JSON). Zero never means unlimited.
|
|
type Limits struct {
|
|
ManifestBytes int64 `json:"manifestBytes"`
|
|
EnvelopeBytes int64 `json:"envelopeBytes"`
|
|
PayloadBytes int64 `json:"payloadBytes"`
|
|
ReadTimeout time.Duration `json:"readTimeout"`
|
|
PreviewTimeout time.Duration `json:"previewTimeout"`
|
|
ApplyTimeout time.Duration `json:"applyTimeout"`
|
|
ShutdownTimeout time.Duration `json:"shutdownTimeout"`
|
|
BackoffInitial time.Duration `json:"backoffInitial"`
|
|
BackoffMax time.Duration `json:"backoffMax"`
|
|
BackoffAttempts int64 `json:"backoffAttempts"`
|
|
BatchSize int64 `json:"batchSize"`
|
|
QueryDepth int64 `json:"queryDepth"`
|
|
QueryRows int64 `json:"queryRows"`
|
|
QueryBytes int64 `json:"queryBytes"`
|
|
ProposalExpiry time.Duration `json:"proposalExpiry"`
|
|
ArtifactExpiry time.Duration `json:"artifactExpiry"`
|
|
ArtifactBytes int64 `json:"artifactBytes"`
|
|
ArtifactQuotaBytes int64 `json:"artifactQuotaBytes"`
|
|
ArtifactCount int64 `json:"artifactCount"`
|
|
Concurrency int64 `json:"concurrency"`
|
|
}
|
|
|
|
func DefaultLimits() Limits {
|
|
return Limits{ManifestBytes: 64 << 10, EnvelopeBytes: 2 << 20, PayloadBytes: 1 << 20, ReadTimeout: 30 * time.Second, PreviewTimeout: 30 * time.Second, ApplyTimeout: 60 * time.Second, ShutdownTimeout: 10 * time.Second, BackoffInitial: 250 * time.Millisecond, BackoffMax: 30 * time.Second, BackoffAttempts: 5, BatchSize: 100, QueryDepth: 8, QueryRows: 1000, QueryBytes: 1 << 20, ProposalExpiry: 5 * time.Minute, ArtifactExpiry: time.Hour, ArtifactBytes: 100 << 20, ArtifactQuotaBytes: 1 << 30, ArtifactCount: 100, Concurrency: 4}
|
|
}
|
|
func (l Limits) Validate() error {
|
|
v := reflect.ValueOf(l)
|
|
for i := 0; i < v.NumField(); i++ {
|
|
if v.Field(i).Int() <= 0 {
|
|
return fmt.Errorf("%s must be positive and finite", v.Type().Field(i).Name)
|
|
}
|
|
}
|
|
if l.BackoffInitial > l.BackoffMax || l.ArtifactBytes > l.ArtifactQuotaBytes || l.PayloadBytes > l.EnvelopeBytes || l.QueryBytes > l.PayloadBytes || l.QueryDepth > 64 {
|
|
return fmt.Errorf("inconsistent limits")
|
|
}
|
|
return nil
|
|
}
|