add the implementation of historyserver collector #4241

KunWuLuan · 2025-12-01T07:04:38Z

add the implementation of historyserver collector

Why are these changes needed?

Related issue number

#3966

Checks

I've made sure the tests are passing.
Testing Strategy
- Unit tests
- Manual tests
- This PR is not tested :(

update go.work go.mod Signed-off-by: KunWuLuan <[email protected]>

Signed-off-by: KunWuLuan <[email protected]>

Future-Outlier · 2025-12-03T16:18:49Z

historyserver/pkg/collector/eventserver/eventserver.go

+
+	// 分类事件
+	for _, event := range eventsToFlush {
+		hourKey := event.Timestamp.Truncate(time.Hour).Format("2006-01-02-15")


should we be "2006-01-02-15" or "2006010215"?

Since this is different from the design doc
ray-project/enhancements#62

machichima · 2025-12-04T12:51:00Z

historyserver/cmd/collector/main.go

+	"github.com/sirupsen/logrus"
+)
+
+const runtimeClassConfigPath = "/var/collector-config/data"


Seems like it's not being used?

machichima · 2025-12-04T13:16:33Z

historyserver/pkg/utils/utils.go

+	for i := 0; i < 12; i++ {
+		rp, err := os.Readlink(session_latest_path)
+		if err != nil {
+			logrus.Errorf("read session_latest file error %v", err)
+			time.Sleep(time.Second * 5)
+			continue
+		}


Looks like we are doing the retry logic here. Could we extract the maxAttempts into constant or something we can configure, and maybe use backoff here?

And curious why we set the max retry to 12 times here? I think it may be too many times for retry, can we use 3?

machichima · 2025-12-04T13:21:01Z

historyserver/pkg/utils/utils.go

+const (
+	RAY_SESSIONDIR_LOGDIR_NAME  = "logs"
+	RAY_SESSIONDIR_METADIR_NAME = "meta"
+)
+
+const (
+	OssMetaFile_BasicInfo = "ack__basicinfo"
+
+	OssMetaFile_NodeSummaryKey                        = "restful__nodes_view_summary"
+	OssMetaFile_Node_Prefix                           = "restful__nodes_"
+	OssMetaFile_JOBTASK_DETAIL_Prefix                 = "restful__api__v0__tasks_detail_job_id_"
+	OssMetaFile_JOBTASK_SUMMARIZE_BY_FUNC_NAME_Prefix = "restful__api__v0__tasks_summarize_by_func_name_job_id_"
+	OssMetaFile_JOBTASK_SUMMARIZE_BY_LINEAGE_Prefix   = "restful__api__v0__tasks_summarize_by_lineage_job_id_"
+	OssMetaFile_JOBDATASETS_Prefix                    = "restful__api__data__datasets_job_id_"
+	OssMetaFile_NodeLogs_Prefix                       = "restful__api__v0__logs_node_id_"
+	OssMetaFile_ClusterStatus                         = "restful__api__cluster_status"
+	OssMetaFile_LOGICAL_ACTORS                        = "restful__logical__actors"
+	OssMetaFile_ALLTASKS_DETAIL                       = "restful__api__v0__tasks_detail"
+	OssMetaFile_Events                                = "restful__events"
+	OssMetaFile_PlacementGroups                       = "restful__api__v0__placement_groups_detail"
+
+	OssMetaFile_ClusterSessionName = "static__api__cluster_session_name"
+
+	OssMetaFile_Jobs         = "restful__api__jobs"
+	OssMetaFile_Applications = "restful__api__serve__applications"
+)


In kuberay we usually put constant in a separate file. For example ray-operator put the constant in ray-operator/controllers/ray/utils/constant.go
Could we move those into a constant.go file?

machichima · 2025-12-04T13:21:44Z

historyserver/pkg/utils/utils.go

+}
+
+func GetSessionDir() (string, error) {
+	session_latest_path := "/tmp/ray/session_latest"


Could we also move this to constant?

Future-Outlier · 2025-12-04T15:36:49Z

historyserver/pkg/collector/logcollector/storage/interface.go

+// StorageWriter is the interface for storage writer.
+type StorageWritter interface {


Can StorageWritter be StorageWriter?

Future-Outlier · 2025-12-04T15:41:59Z

historyserver/pkg/collector/types/types.go

+	PushInterval   time.Duration
+}


Where is PushInterval be used?

Future-Outlier · 2025-12-04T16:02:29Z

historyserver/pkg/collector/types/types.go

+	Role           string
+	RayClusterName string
+	RayClusterID   string
+	LogBatching    int


Where is this be used?

fscnick · 2025-12-06T03:43:11Z

historyserver/Dockerfile

+ARG TARGETARCH
+
+FROM --platform=$BUILDPLATFORM golang:1.25.1 as builder
+ENV GOPROXY=https://goproxy.cn,direct


Should it be the official proxy? https://proxy.golang.org/

ref: https://proxy.golang.org/

fscnick · 2025-12-06T06:47:44Z

historyserver/Dockerfile

+
+RUN apt-get update && apt-get upgrade -y && rm -rf /var/cache/apt/ && apt-get install -y ca-certificates
+
+COPY --from=builder /historyserver/output/bin/historyserver /usr/local/bin/historyserver


The historyserver is not ready in this pr. Should it be removed?

fscnick · 2025-12-06T06:49:11Z

historyserver/Dockerfile

+COPY --from=builder /historyserver/dashboard/v2.51.0/client/build /dashboard/v2.51.0/client/build
+COPY --from=builder /historyserver/dashboard/homepage /dashboard/homepage


Should these COPY consider BUILD_RAYSERVER_DASHBOARD? Or, it might be failed on building.

fscnick · 2025-12-06T06:52:25Z

historyserver/Makefile

+	rm -rf $(OUT_DIR)
+
+.PHONY: build
+build: buildcollector buildhistoryserver


should the buildhistoryserver be removed because it is not ready in this pr? And so does the historyserver related content in this file.

fscnick · 2025-12-06T07:03:56Z

historyserver/cmd/collector/main.go

+	flag.StringVar(&role, "role", "Worker", "")
+	flag.StringVar(&runtimeClassName, "runtime-class-name", "", "")
+	flag.StringVar(&rayClusterName, "ray-cluster-name", "", "")
+	flag.StringVar(&rayClusterId, "ray-cluster-id", "default", "")
+	flag.StringVar(&rayRootDir, "ray-root-dir", "", "")
+	flag.IntVar(&logBatching, "log-batching", 1000, "")
+	flag.IntVar(&eventsPort, "events-port", 8080, "")
+	flag.StringVar(&runtimeClassConfigPath, "runtime-class-config-path", "", "") //"/var/collector-config/data"
+	flag.DurationVar(&pushInterval, "push-interval", time.Minute, "")


It might be better to have some usage description.

add the implementation of historyserver collector

091caec

update go.work go.mod Signed-off-by: KunWuLuan <[email protected]>

Future-Outlier self-assigned this Dec 1, 2025

KunWuLuan mentioned this pull request Dec 1, 2025

[Feature] Support History Server #3966

Open

2 tasks

update the func judging if the event is releated to the Nodes.

7b2acae

Signed-off-by: KunWuLuan <[email protected]>

Future-Outlier reviewed Dec 3, 2025

View reviewed changes

machichima reviewed Dec 4, 2025

View reviewed changes

Future-Outlier reviewed Dec 4, 2025

View reviewed changes

fscnick reviewed Dec 6, 2025

View reviewed changes

		// StorageWriter is the interface for storage writer.
		type StorageWritter interface {


		RUN apt-get update && apt-get upgrade -y && rm -rf /var/cache/apt/ && apt-get install -y ca-certificates

		COPY --from=builder /historyserver/output/bin/historyserver /usr/local/bin/historyserver

		COPY --from=builder /historyserver/dashboard/v2.51.0/client/build /dashboard/v2.51.0/client/build
		COPY --from=builder /historyserver/dashboard/homepage /dashboard/homepage No newline at end of file

add the implementation of historyserver collector #4241

Are you sure you want to change the base?

add the implementation of historyserver collector #4241

Uh oh!

Conversation

KunWuLuan commented Dec 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Why are these changes needed?

Related issue number

Checks

Uh oh!

Future-Outlier Dec 3, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

KunWuLuan commented Dec 1, 2025 •

edited

Loading

Future-Outlier Dec 3, 2025 •

edited

Loading