ryerraguntla commented on code in PR #4043: URL: https://github.com/apache/iggy/pull/4043#discussion_r4001397302
########## gateways/kafka/src/bridge/iggy_bridge.rs: ########## @@ -0,0 +1,381 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +use std::time::Duration; + +use iggy::prelude::{ + AutoLogin, Client, Credentials, Identifier, IggyClient, IggyClientBuilder, IggyError, + StreamClient, TopicClient, TopicCreateOptions, +}; +use tracing::{debug, info}; + +use crate::bridge::config::IggyBridgeConfig; +use crate::bridge::error::BridgeError; + +/// Passes attempted, after the first, before [`IggyBridge::connect`] gives up and returns `Err`. +/// +/// Not the SDK's own default (`TcpClientReconnectionConfig::default()` is `max_retries: None` - +/// unlimited, one dial per second, forever). A Kafka client already retries at the wire-protocol +/// level once a handler maps a bridge failure to a retriable error code; the bridge blocking a +/// request task inside an unbounded internal reconnect loop would just add a second, invisible +/// retry layer underneath that one instead of surfacing the failure so the mapped code can be +/// sent. +/// +/// This bounds the *count*, not the *wall-clock time*, of that inner retry loop - see +/// [`CONNECT_TIMEOUT`] for the latter. +const RECONNECTION_RETRIES: u32 = 3; + +/// Wall-clock ceiling on the whole `client.connect()` call in [`IggyBridge::connect`], including +/// every attempt [`RECONNECTION_RETRIES`] makes internally. +/// +/// Without this, an unreachable-but-not-refusing address hangs far longer than "a few seconds": +/// `TcpClient::establish_bounded` only applies its own `FAILOVER_DIAL_TIMEOUT` (2s) when at least +/// two failover candidates are configured (`tcp_client.rs`) - a bridge always configures exactly +/// one address, so that guard never engages, and the plain `TcpStream::connect` underneath has no +/// deadline of its own. Against a firewall that drops SYN packets instead of refusing them, each +/// of the up to `RECONNECTION_RETRIES + 1` dial attempts pays the kernel's own SYN-retry timeout +/// (minutes, not seconds) rather than the `reconnection_interval` between attempts - a closed port +/// (instant RST) never exercises this path, so the failure mode only shows up in production. +/// 15s comfortably covers a slow-but-alive server's handshake (well above p99 login latency) while +/// still failing well short of the pathological multi-minute case. +const CONNECT_TIMEOUT: Duration = Duration::from_secs(15); + +/// Owns one connected `IggyClient` and resolves Kafka topics against it. +/// +/// Produce/Fetch handler wiring is a separate, later change (`#3535`/`#3536`) - this type is the +/// shared plumbing those handlers will call into, exercised standalone here via its own tests and +/// an integration test against a real `iggy-server`. +pub struct IggyBridge { + client: IggyClient, Review Comment: **Fixed** (documented, not pooled). Went with documenting per your suggested fallback: IggyBridge's own doc comment now states the lockstep-transport ceiling directly (citing tcp_client.rs's own "lockstep" comment), and the README gained a "Concurrency ceiling" section saying the same for operators. Pooling is flagged as a known gap to close before #3535/#3536, not attempted here. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
