Fix dropping JVM metrics when collecting multiple skywalking instance…

…s's data (#1163)

* fix missing skywalking metrics issue

* add changelog

* update changelog

---------

Co-authored-by: Tom Yu <[email protected]>

CHANGELOG.md

@@ -36,11 +36,13 @@ your changes, such as:
 - [public] [both] [updated] add a new feature
 
 ## [Unreleased]
+- [public] [both] [fixed] fix dropping jvm metrics when collecting multiple skywalking instances's data
+- [public] [both] [fixed] fix elasticsearch flusher authentication tls config and http config
 - [public] [both] [fixed] fix profiling wrong type when the different profiling type having same stack. 
 - [public] [both] [added] add UsingOldContentTag. When UsingOldContentTag is set to false, the Tag is now placed in the Meta instead of Logs during cgo.
 - [public] [both] [fixed] fix send local buffer failed when upgrade iLogtail from version earlier than 1.3.
 - [public] [both] [updated] Updated strptime_ns to parse %c format from "%x %X" to "%a %b %d %H:%M:%S %Y" for consistent behavior with striptime.
 - [public] [both] [fixed] fix topic key does not support underscore.
 - [public] [both] [fixed] fix jmxfetch status error when exist multi jmxfetch config in the same machine.
-- [public] [both] [fixed] fix user container wss increase problem.
+- [public] [both] [fixed] fix increasing WSS memory issue in collected containers.
 - [public] [both] [fixed] fix cannot log blacklist config error

plugins/input/skywalkingv2/jvm_metric_handle.go

@@ -18,6 +18,7 @@ import (
 	"context"
 
 	"github.com/alibaba/ilogtail/pkg/helper"
+	"github.com/alibaba/ilogtail/pkg/logger"
 	"github.com/alibaba/ilogtail/pkg/pipeline"
 	"github.com/alibaba/ilogtail/pkg/protocol"
 	"github.com/alibaba/ilogtail/plugins/input/skywalkingv2/skywalking/apm/network/language/agent"
@@ -35,6 +36,7 @@ func (j *JVMMetricServiceHandle) Collect(ctx context.Context, metrics *agent.JVM
 	applicationInstance, ok := j.RegistryInformationCache.findApplicationInstanceRegistryInfo(metrics.ApplicationInstanceId)
 
 	if !ok {
+		logger.Warning(j.context.GetRuntimeContext(), "Failed to find application instance registry info", "applicationInstanceId", metrics.ApplicationInstanceId)
 		return &agent.Downstream{}, nil
 	}
 

plugins/input/skywalkingv3/jvm_metric_handle.go

@@ -34,13 +34,15 @@ type JVMMetricHandler struct {
 func (h *JVMMetricHandler) Collect(ctx context.Context, req *skywalking.JVMMetricCollection) (*v3.Commands, error) {
 	defer panicRecover()
 	for _, metric := range req.GetMetrics() {
-		fixTime := metric.Time - metric.Time%1000
-		// 原始数据上传过于频繁, 不需要
-		if fixTime-h.lastTime >= h.interval {
-			h.lastTime = fixTime
-		} else {
-			return &v3.Commands{}, nil
-		}
+		// 在充当Collector角色（一对多个实例）时，将会出现问题，因为每个实例的时间不一致，所以这里不做时间过滤
+		// fixTime := metric.Time - metric.Time%1000
+		// // 原始数据上传过于频繁, 不需要
+		// if fixTime-h.lastTime >= h.interval {
+		// 	h.lastTime = fixTime
+		// } else {
+		// 	logger.Warning(h.context.GetRuntimeContext(), "", "lastTime", h.lastTime, "fixTime", fixTime, "interval", h.interval)
+		// 	return &v3.Commands{}, nil
+		// }
 		logs := h.toMetricStoreFormat(metric, req.Service, req.ServiceInstance)
 		for _, log := range logs {
 			h.collector.AddRawLog(log)

plugins/input/skywalkingv3/management_handle.go

@@ -79,8 +79,6 @@ func (r *ResourcePropertiesCache) save(ctx pipeline.Context) {
 	err := ctx.SaveCheckPoint(r.cacheKey, jsonBytes)
 	if err != nil {
 		logger.Error(ctx.GetRuntimeContext(), "SKYWALKING_SAVE_CHECKPOINT_FAIL", "err", err.Error())
-	} else {
-		logger.Info(ctx.GetRuntimeContext(), "msg", "skywalking save checkpoint done")
 	}
 }
 
@@ -92,7 +90,6 @@ func (r *ResourcePropertiesCache) load(ctx pipeline.Context) bool {
 			logger.Error(ctx.GetRuntimeContext(), "SKYWALKING_LOAD_CHECKPOINT_FAIL", "err", err.Error())
 			return false
 		}
-		logger.Info(ctx.GetRuntimeContext(), "msg", "skywalking load checkpoint done", "count", len(r.cache))
 	}
 	return true
 }