This document was translated by GPT-4

1. Universal Service Map, Application, Networks, Infrastructure, Events

• AutoMetrics
  • Supports parsing of FastCGI Protocol
  • Supports parsing of HengSheng T3 Protocol
  • Supports parsing of event and serialization_id fields for Dubbo Protocol
  • Recognizes RST disconnection in SLB detection as normal behavior
  • Adds an endpoint field to the Application Aggregate Metrics table
  • Supports collection of statement-ids for MySQL calls to associate COM_STMT_EXECUTE with COM_STMT_QUERY and trace SQL statements
• AutoTracing
  • Supports parsing of Tingyun's Tracing field X-Tingyun
  • Supports distributed tracing of calls before and after ALB/SLB services
  • Supports parsing of TraceID in MySQL statements
  • Adds allow_multiple_trace_ids_in_tracing_result configuration item to allow multiple TraceIDs in the distributed tracing result
  • Supports enhancing distributed tracing data by calling APM's Trace API
• AutoTagging
  • Supports automatic association of K8s Annotation and Env tags
  • Adds the ability to associate K8s containers automatically by PID to resolve the tagging issue of container resource information in HostNetwork Pod in eBPF data
• SQL
  • Adds count operator to metrics to calculate the number of rows in the original data
• Universal Service Map
  • Optimizes the service list
    • Only shows performance metrics when the service is called
    • Supports switching the signal source through signal_source
    • Adjusts the default display metrics and shows a bar graph representing the relative size of the metrics
  • Adds service topology
    • Supports defining an integrated business access topology for both cloud and non-cloud environments
• Application
  • Distributed Tracing
    • When adjusting the split ratio of the page by sliding right, do not scale the flame graph
    • Supports distributed tracing from the network Span as a starting point
    • Supports showing the collection network card information of the network Span in the flame graph
    • The quick filter box on the left supports quick filtering and switching of the signal source
  • Continuous Profiling
    • Supports the linked display of tables and flame graphs
    • Optimizes the merging logic of Function Stack in the flame graph
    • Compresses the storage of Function Stack in ClickHouse
    • Support for eBPF collection of OnCPU Profile data for compiled (Golang/Rust, etc.) and interpreted (Java, etc.) languages
• Infrastructure
  • Adds Host and Container pages, displayed based on Prometheus metrics
• Events
  • Splits the Event page into Resource Change, File Read/Write, and Alert events pages
• GUI
  • Enhances snapshot search capabilities: supports sorting, condition copying, and other operations
  • Search bar supports pasting Key: value search conditions
  • Search bar supports modifying the operator of existing conditions
  • Simplifies the search conditions of the page and the conditions carried by the right-sliding page, improving usability
  • Comprehensive optimization of page UI
  • Optimizes the Quick Filter box on the left: supports search filtering, displays matched data quantity, supports filtering of metric value range, supports switching query region, supports switching data table
  • Supports opening multiple right-sliding pages in succession and switching between different right-sliding pages
  • The event data displayed in the right-sliding page can switch to view the events of the client or server separately
  • The pop-up page for viewing the database field supports displaying the table name
  • Supports viewing the search conditions of the current subview

2. Views, Metrics, Alerts, Reports

• Views
  • Supports dragging to modify the table size
  • Optimizes the display details of the bar chart
  • Adds a new subview type: Overview Diagram
  • Refactoring of the subview editing page
  • Optimizes the layout of subview buttons
  • Merges Line Chart with the Top Line Chart
  • Optimizes the display when template variable names are too long
  • Supports dragging to order the template variable list
• Metrics
  • Supports inputting PromQL to query data
• Alerts
  • Supports creating alert policies directly (no need to create views)
  • HTTP Push Endpoint supports using tags to render push content
  • Email Push title supports using variables
  • Optimizes the display of system alert events
• GUI
  • Unifies search condition input boxes of Subview Editing Page, Metric Search Page, and Alert Policy Editing Page

3. Resources, Systems

• Resources
  • Optimizes display of POD list, VPC list, availability zone list, region list, adds container node quantity, collector status column
  • Supports sync of OpenKruise CloneSet and Advanced StatefulSet workload types
  • Support independent configuration of sync intervals for different cloud platforms
  • Supports sync of IP address on loopback interface (usually VIP)
• Integration
  • Prometheus Integration
    • PromQL supports topk, bottomk functions
    • PromQL API supports RFC3339 time format
    • Supports obtaining HTTP Header in RemoteWrite as an additional Label
    • Optimizes the storage performance of RemoteWrite, and optimizes query performance of RemoteRead and PromQL
  • OpenTelemetry Integration
    • Supports operation without ClickHouse
• Agent
  • Plugin
    • New "so plugin" support capability, providing C SDK
    • Wasm Demo: Parses error code in HTTP payload, and reassigns response_code and response_exception
    • Wasm Demo: Parses Protobuf message in Payload
  • The cycle reporting interval of long-lived streams is changed from absolute 0 seconds (at the beginning of every minute) to relative 0 seconds (in relation to the start of the stream every 60 minutes)
    • Advantage: Reduces pressure of sending stream logs at absolute 0 seconds, avoid splitting streams that last less than 60 seconds into two stream logs
  • Configuration
    • Supports configuration of CPU affinity and priority
    • Adds kprobe-blacklist configuration item, sets the port number blacklist for eBPF data collection, avoiding collection loops
    • Adds l4_log_ignore_tap_sides and l7_log_ignore_tap_sides for setting positions to ignore in flow logs and invocation logs, respectively, to reduce the amount of data collected
  • Adaptation
    • Supports operation in Tencent TCE's DPDK host machine
    • Removes HostNetwork requirement for container collector
    • Supports environments where the number of matched network card (tap_interface_regex) results exceeds 255
    • Supports running as a Sidecar within a business Pod
    • Supports deployment as a blue whale plugin
• Server
  • Supports configuration of system alert sending mailbox on the page
  • The default value for the auto-deletion time of the anomaly controller and data nodes is set to 30 days
  • Unifies storage of alert events in ClickHouse
  • Detects changes in resource information, and pushes updates to the Agent in real time
  • Supports disabling of K8s cluster auto-discovery function, so it can be used as a synced K8s cluster for public cloud
  • Supports specifying (fixing) the Agent that syncs K8s resource information
  • Adds deepflow identifier to all deployed dependent HostPath paths
• CLI
  • Releases deepflow-ctl for MacOS