Skip to content

feat: Add retry mechanism to Ollama and OpenAiCompatible providers #114

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

Merged
kaspermarstal merged 1 commit into from
Feb 18, 2025
Merged

feat: Add retry mechanism to Ollama and OpenAiCompatible providers #114

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
3 changes: 2 additions & 1 deletion src/Cellm.Models/Providers/OpenAiCompatible/OpenAiCompatibleRequestHandler.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,11 +2,12 @@
using System.ClientModel.Primitives;
using Cellm.Models.Prompts;
using Microsoft.Extensions.AI;
using Microsoft.Extensions.DependencyInjection;
using OpenAI;

namespace Cellm.Models.Providers.OpenAiCompatible;

internal class OpenAiCompatibleRequestHandler(HttpClient httpClient)
internal class OpenAiCompatibleRequestHandler([FromKeyedServices("ResilientHttpClient")] HttpClient httpClient)
: IModelRequestHandler<OpenAiCompatibleRequest, OpenAiCompatibleResponse>
{

Expand Down
30 changes: 23 additions & 7 deletions src/Cellm.Models/ServiceCollectionExtensions.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,9 +25,11 @@ public static IServiceCollection AddAnthropicChatClient(this IServiceCollection
{
anthropicHttpClient.BaseAddress = anthropicConfiguration.BaseAddress;
anthropicHttpClient.DefaultRequestHeaders.Add("anthropic-version", anthropicConfiguration.Version);
anthropicHttpClient.Timeout = TimeSpan.FromHours(1);
anthropicHttpClient.Timeout = TimeSpan.FromSeconds(configuration
.GetSection(nameof(ProviderConfiguration))
.GetValue<int>(nameof(ProviderConfiguration.HttpTimeoutInSeconds)));
})
.AddResilienceHandler($"{nameof(AnthropicRequestHandler)}", resiliencePipelineConfigurator.ConfigureResiliencePipeline);
.AddResilienceHandler($"{nameof(AnthropicRequestHandler)}{nameof(ResiliencePipelineConfigurator)}", resiliencePipelineConfigurator.ConfigureResiliencePipeline);

// TODO: Add IChatClient-compatible Anthropic client

Expand All @@ -36,20 +38,36 @@ public static IServiceCollection AddAnthropicChatClient(this IServiceCollection

public static IServiceCollection AddOllamaChatClient(this IServiceCollection services, IConfiguration configuration)
{
var resiliencePipelineConfigurator = new ResiliencePipelineConfigurator(configuration);

var ollamaConfiguration = configuration.GetRequiredSection(nameof(OllamaConfiguration)).Get<OllamaConfiguration>()
?? throw new NullReferenceException(nameof(OllamaConfiguration));

services
.AddKeyedChatClient(Provider.Ollama, serviceProvider => new OllamaChatClient(
ollamaConfiguration.BaseAddress,
ollamaConfiguration.DefaultModel))
ollamaConfiguration.DefaultModel,
serviceProvider.GetKeyedService<HttpClient>("ResilientHttpClient")))
.UseFunctionInvocation();

return services;
}

public static IServiceCollection AddResilientHttpClient(this IServiceCollection services, IConfiguration configuration)
{
var resiliencePipelineConfigurator = new ResiliencePipelineConfigurator(configuration);

services
.AddHttpClient("ResilientHttpClient", resilientHttpClient =>
{
resilientHttpClient.Timeout = TimeSpan.FromSeconds(configuration
.GetSection(nameof(ProviderConfiguration))
.GetValue<int>(nameof(ProviderConfiguration.HttpTimeoutInSeconds)));
})
.AddAsKeyed()
.AddResilienceHandler("ResilientHttpClientHandler", resiliencePipelineConfigurator.ConfigureResiliencePipeline);

return services;
}

public static IServiceCollection AddSentryBehavior(this IServiceCollection services)
{
services
Expand Down Expand Up @@ -85,8 +103,6 @@ public static IServiceCollection AddTools(this IServiceCollection services, para

public static IServiceCollection AddTools(this IServiceCollection services, params Func<IServiceProvider, AIFunction>[] toolBuilders)
{


foreach (var toolBuilder in toolBuilders)
{
services.AddSingleton((serviceProvider) => toolBuilder(serviceProvider));
Expand Down
12 changes: 2 additions & 10 deletions src/Cellm/Services/ServiceLocator.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,15 +104,8 @@ private static IServiceCollection ConfigureServices(IServiceCollection services)
// Add providers
services
.AddAnthropicChatClient(configuration)
.AddOllamaChatClient(configuration);

services
.AddHttpClient<IModelRequestHandler<OpenAiCompatibleRequest, OpenAiCompatibleResponse>>(openAiCompatibleHttpClient =>
{
openAiCompatibleHttpClient.Timeout = TimeSpan.FromSeconds(configuration
.GetSection(nameof(ProviderConfiguration))
.GetValue<int>(nameof(ProviderConfiguration.HttpTimeoutInSeconds)));
});
.AddOllamaChatClient(configuration)
.AddResilientHttpClient(configuration);

// Add provider middleware
services
Expand All @@ -124,7 +117,6 @@ private static IServiceCollection ConfigureServices(IServiceCollection services)
.AddSingleton<FileReaderFactory>()
.AddSingleton<IFileReader, PdfReader>()
.AddSingleton<IFileReader, TextReader>()
.AddSingleton<Functions>()
.AddTools(
serviceProvider => AIFunctionFactory.Create(serviceProvider.GetRequiredService<Functions>().GlobRequest),
serviceProvider => AIFunctionFactory.Create(serviceProvider.GetRequiredService<Functions>().FileReaderRequest));
Expand Down
1 change: 0 additions & 1 deletion src/Cellm/appsettings.Local.Llamafile.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,5 @@
{
"ProviderConfiguration": {
"BaseAddress": "http://127.0.0.1:8080",
"DefaultProvider": "Llamafile"
}
}
6 changes: 3 additions & 3 deletions src/Cellm/appsettings.Local.OpenAiCompatible.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,8 +1,8 @@
{
"OpenAiCompatibleConfiguration": {
"BaseAddress": "YOUR_API_ENDPOINT",
"DefaultModel": "YOUR_DEFAULT_MODEL",
"ApiKey": "YOUR_API_KEY_OPTIONAL"
"BaseAddress": "BASE_ADDRESS",
"DefaultModel": "MODEL_NAME",
"ApiKey": "API_KEY"
},
"ProviderConfiguration": {
"DefaultProvider": "OpenAiCompatible"
Expand Down
18 changes: 9 additions & 9 deletions src/Cellm/appsettings.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,28 +72,28 @@
"DefaultModel": "gemma2:2b",
"DefaultTemperature": 0,
"CacheTimeoutInSeconds": 3600,
"HttpTimeoutInSeconds": 3600,
"HttpTimeoutInSeconds": 900,
"EnableCache": true,
"EnableTools": {
"GlobRequest": false,
"FileReaderRequest": false
}
},
"RateLimiterConfiguration": {
"TokenLimit": 4,
"QueueLimit": 3600,
"ConcurrencyLimit": 4,
"ReplenishmentPeriodInSeconds": 2,
"TokenLimit": 2,
"QueueLimit": 16384,
"ConcurrencyLimit": 2,
"ReplenishmentPeriodInSeconds": 1,
"TokensPerPeriod": 1
},
"CircuitBreakerConfiguration": {
"FailureRatio": 0.2,
"FailureRatio": 0.3,
"SamplingDurationInSeconds": 30,
"MinimumThroughput": 10,
"BreakDurationInSeconds": 4
"MinimumThroughput": 4,
"BreakDurationInSeconds": 8
},
"RetryConfiguration": {
"MaxRetryAttempts": 4,
"MaxRetryAttempts": 3,
"DelayInSeconds": 4
},
"SentryConfiguration": {
Expand Down