From 6b7ddbbe92c91f8098c738ef9b809a68651fbfb0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sunxien=C2=AE?= <81640257+sunxien@users.noreply.github.com>
Date: Fri, 16 May 2025 15:09:22 +0800
Subject: [PATCH] =?UTF-8?q?fix(canal/prometheus):=20=E4=BF=AE=E5=A4=8D?=
 =?UTF-8?q?=E5=BB=B6=E8=BF=9F=E6=8C=87=E6=A0=87=E7=9A=84=E7=BC=BA=E9=99=B7?=
 =?UTF-8?q?=20(#5426)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

-【缺陷描述】
   Admin控制台修改canal.properties配置后，Server自动进行重启，在Grafan上观测到同步实例的PUT/GET/ACK延迟居高不下。

-【原因定位】
   源库心跳正常触发，MemoryStoreWithBuffer正常推进位点，profiling 正常统计，但是延迟指标依然是越来越来越高。Prometheus调用collect接口采集到的exec time时间始终是固定不变的。经debug排查到StoreCollector采集器内存hold的StoreMetricsHolder与CanalInstance实例中的引用已经不同啦，CanalIntance重启时已经被重建过一份新的实例，但是StoreMetricsHolder却没有保存到内存Hold中。原因是Map.putIfAbsent调用引起。

-【修复效果】
- 修复后，重启Server心跳正常推进，延迟瞬间降下来。梳理其他Collector代码都是调用Map.put，只有这里使用putIfAbsent可能是粗心导致吧。修复效果
---
 .../com/alibaba/otter/canal/prometheus/impl/StoreCollector.java | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/prometheus/src/main/java/com/alibaba/otter/canal/prometheus/impl/StoreCollector.java b/prometheus/src/main/java/com/alibaba/otter/canal/prometheus/impl/StoreCollector.java
index e22eb251..81b90ba7 100644
--- a/prometheus/src/main/java/com/alibaba/otter/canal/prometheus/impl/StoreCollector.java
+++ b/prometheus/src/main/java/com/alibaba/otter/canal/prometheus/impl/StoreCollector.java
@@ -162,7 +162,7 @@ public class StoreCollector extends Collector implements InstanceRegistry {
             Preconditions.checkNotNull(holder.putMemSize);
             Preconditions.checkNotNull(holder.ackMemSize);
         }
-        StoreMetricsHolder old = instances.putIfAbsent(destination, holder);
+        StoreMetricsHolder old = instances.put(destination, holder);
         if (old != null) {
             logger.warn("Remote stale StoreCollector for instance {}.", destination);
         }