Skip to content

Releases: databufflabs/databuff

Release list

v0.1.4 — AI Ops Capability Home + Multi-Agent + Data Accuracy Hardening

v0.1.4 — AI Ops Capability Home + Multi-Agent + Data Accuracy Hardening Latest
Latest

Choose a tag to compare

View all tags
@databufflabs databufflabs released this 21 Jul 05:43
v0.1.4
74d7bba

Highlights

  • AI Ops capability home — Seven configurable AI capabilities on the chat home: observe, multi-agent, inspect, diagnose, repair, predict, answer. Capability name, suggested questions, and bound expert are all configurable; enable/disable or reset to defaults from the admin page.
  • Multi-agent collaboration — Different experts run in parallel, same expert serial, so context and execution state no longer clobber each other. Task / session / generated-file ownership fixed; HTML / Markdown / SVG reports preview and download from the chat workspace.
  • Image understanding & file workspace — Upload PNG / JPEG / GIF / WebP screenshots into the analysis; workspace carries generated files end-to-end from screenshot → analysis → report.
  • Product Q&A expert — Retrieves DataBuff docs and code inside the image, then answers with verification steps instead of guessing from model memory.
  • Deep service inspection — One sentence kicks off an inspection across entry metrics, logs, active alerts, upstream/downstream deps, error traces, instances, and JVM / GC / CPU / memory signals; outputs an HTML report with anomaly location, evidence, and remediation.
  • SkyWalking / Trace data accuracy — Adds MQ virtual services; corrects RPC CLIENT / outbound peer relations, entry-service stats, and true max latency; traces aligned by end-minute, fixing execution-ratio and end-time semantics.
  • Default entry alerts — Seeds three entry alerts (avg latency, error rate, exception count); fixes instant-alert counting for zero-duration alerts.
  • Long-write pipeline — Log body as STRING; long URL / SQL / log and multibyte-CN truncation boundaries; poison-batch fail-soft so one bad batch no longer blocks the whole table.
  • Doris runtime recovery — When Doris is down, Web keeps the AI troubleshooting entry; on recovery it auto-detects and hydrates persisted state without a Web restart.
  • Legacy Docker Compose compat — docker-compose.legacy.yml covers docker-compose v1.18–1.28; start.sh / update.sh auto-select it.
  • Upgrade guardian — update.sh self-copies to temp then re-exec; backs up data/ before upgrade, auto-restores and retries on failure (3x); manifest schema CI guard added.
  • AI Shell guardrails — Default dangerous-command denylist, extendable via apm.agent.shell-command-denylist; denylist is a default rail, not a full sandbox.
  • AI session bounds — In-memory sessions get a 1-day TTL and 500-entry cap, LRU-evicting idle sessions; running sessions are never evicted.

Quick Start

# Platform (latest = 0.1.4)
curl -fsSL https://databuff.ai/databuff/ai-apm-install.sh | bash

# Upgrade existing Docker install (keeps data/)
curl -fsSL https://databuff.ai/databuff/ai-apm-update.sh | bash

Open http://YOUR_HOST:27403 · login admin / Databuff@123 · configure LLM API key for AI features.

Offline Install

tar -zxvf databuff-ai-apm-offline-0.1.4-amd64.tar.gz
cd databuff-ai-apm-offline-0.1.4-amd64
sudo ./update.sh

Links

Full changelog

  • 7 AI Ops capabilities on chat home, configurable name / suggested questions / bound expert
  • Multi-agent orchestration: parallel across experts, serial within an expert; fixed task/session/file ownership; HTML/MD/SVG report preview & download
  • Image understanding (PNG/JPEG/GIF/WebP) and file workspace
  • Product Q&A expert retrieving in-image DataBuff docs and code
  • Deep service inspection producing HTML reports with evidence and remediation
  • SkyWalking MQ virtual services; RPC CLIENT/outbound peer, entry stats, true max latency fixes
  • Trace aligned by end-minute; execution-ratio and end-time semantics fixed
  • Three default entry alerts (avg latency / error rate / exception count); instant-alert count fix
  • Long-write pipeline: log body as STRING, long URL/SQL/log + multibyte-CN truncation, poison-batch fail-soft
  • Doris runtime recovery: Web keeps AI entry, auto-detects and hydrates state on Doris return, no Web restart
  • Legacy Docker Compose compat via docker-compose.legacy.yml (v1.18–1.28)
  • Upgrade guardian: temp self-copy + re-exec, data/ backup, auto-restore + retry (3x), manifest schema CI guard
  • AI Shell dangerous-command denylist, extendable via apm.agent.shell-command-denylist
  • AI in-memory sessions: 1-day TTL, 500-entry cap, LRU eviction of idle sessions
Assets 2
Loading

v0.1.3 — SkyWalking Native Ingest + Ops Expert SSH Troubleshooting

v0.1.3 — SkyWalking Native Ingest + Ops Expert SSH Troubleshooting

Choose a tag to compare

View all tags
@databufflabs databufflabs released this 13 Jul 08:59
0.1.3
a611976

Highlights

  • SkyWalking native gRPC ingest — Ingest listens on 11800 (same default port as SkyWalking OAP). Keep your existing Java Agent jar; point collector.backend_service at DataBuff to get traces, JVM metrics, topology, and log correlation on the same platform
  • Ops Expert · install recovery — When install.sh / start.sh exits non-zero but Doris is not ready, the Web UI still boots in troubleshooting mode: configure your LLM API key, pick the built-in Ops Expert (expertId=ops), and SSH read-only to inspect containers, Doris logs, and cgroup limits
  • Ops Expert · live runtime troubleshooting — When checkout P99 spikes but the root cause is unclear (JVM vs host vs downstream), Ops Expert runs read-only Bash/SSH on the target host and aligns findings with slow Trace spans
  • Docker in-place upgrade — ai-apm-update.sh upgrades to 0.1.3 without wiping data/
  • Offline install bundles — Full offline packages per architecture (databuff-ai-apm-offline-0.1.3-<arch>.tar.gz)
  • Live demo — https://demo.databuff.ai (admin / Databuff@123)
  • Bash tool for AI experts — Built-in experts can execute read-only shell commands for deployment and runtime checks

Quick Start

# Platform (latest = 0.1.3)
curl -fsSL https://databuff.ai/databuff/ai-apm-install.sh | bash

# Upgrade existing Docker install (keeps data/)
curl -fsSL https://databuff.ai/databuff/ai-apm-update.sh | bash

# Demo apps (optional)
curl -fsSL https://databuff.ai/databuff/ai-apm-demo-install.sh | bash

Open http://YOUR_HOST:27403 · login admin / Databuff@123 · configure LLM API key for AI features.

Already running SkyWalking Agents? Point collector.backend_service at your DataBuff ingest host on port 11800.

Offline Install

# Example: amd64 offline bundle
tar -zxvf databuff-ai-apm-offline-0.1.3-amd64.tar.gz
cd databuff-ai-apm-offline-0.1.3-amd64
sudo ./install.sh

Kubernetes

curl -fsSL https://databuff.ai/databuff/ai-apm-k8s-install.sh | bash
curl -fsSL https://databuff.ai/databuff/ai-apm-demo-k8s-install.sh | bash

What's included

Component Image
ai-apm-web databuffhub/ai-apm-web:0.1.3
ai-apm-ingest databuffhub/ai-apm-ingest:0.1.3
ai-apm-demo databuffhub/ai-apm-demo:0.1.3

Demo

DataBuff demo

Links

Full changelog

  • SkyWalking native gRPC ingest on port 11800 with demo seeder support
  • Built-in Ops Expert for install recovery and live SSH troubleshooting
  • Web troubleshooting bootstrap when Doris is down (LLM setup + SSH triage)
  • Hardened health checks, JDBC fast-fail, and Doris failover E2E fixes
  • Fix 5s API hang when Doris is unreachable (fail-close JDBC gate before web port opens)
  • Upgrade script retries up to three times on transient failures
  • SkyWalking db.type/SQL normalization and Dubbo ingest fixes
  • Service flow entry service bug fix; trace detail span color classification
  • AI session count uses full DB total; history drawer infinite scroll
  • 30-day dynamic partition retention on Doris metric tables
  • CONTRIBUTING.md, Issue templates, and community links in README
  • Docker env example and expanded ops docs
Loading

v0.1.2

v0.1.2

Choose a tag to compare

View all tags
@databufflabs databufflabs released this 06 Jul 14:15
0.1.2
86ad810

Highlights

  • Docker in-place upgrade — ai-apm-update.sh upgrades without wiping data/; includes schema migrations and verification
  • Log analysis — Ingest and analyze application logs alongside traces and metrics
  • Offline install bundles — Full offline packages per architecture (databuff-ai-apm-offline-0.1.2-<arch>.tar.gz)
  • MCP & Skills — Expose platform capabilities to Cursor / Claude; ship example skills for metrics and health inspection
  • Live demo — https://demo.databuff.ai (admin / Databuff@123)
  • API key masking — Sensitive keys are desensitized in the UI
  • OpenTelemetry OTLP docs — Clear ingestion guide for traces, metrics, and logs

Quick Start

# Platform (latest = 0.1.2)
curl -fsSL https://databuff.ai/databuff/ai-apm-install.sh | bash

# Upgrade existing Docker install (keeps data/)
curl -fsSL https://databuff.ai/databuff/ai-apm-update.sh | bash

# Demo apps (optional)
curl -fsSL https://databuff.ai/databuff/ai-apm-demo-install.sh | bash

Open http://YOUR_HOST:27403 · login admin / Databuff@123 · configure LLM API key for AI features.

Offline Install

# Example: amd64 offline bundle
tar -zxvf databuff-ai-apm-offline-0.1.2-amd64.tar.gz
cd databuff-ai-apm-offline-0.1.2-amd64
sudo ./install.sh

Kubernetes

curl -fsSL https://databuff.ai/databuff/ai-apm-k8s-install.sh | bash
curl -fsSL https://databuff.ai/databuff/ai-apm-demo-k8s-install.sh | bash

What's included

Component Image
ai-apm-web databuffhub/ai-apm-web:0.1.2
ai-apm-ingest databuffhub/ai-apm-ingest:0.1.2
ai-apm-demo databuffhub/ai-apm-demo:0.1.2

Demo

DataBuff demo

Links

Full changelog

  • Docker in-place upgrade with schema migrations and verification
  • Log ingestion and log analysis in AI platform
  • Offline download and offline install bundles
  • MCP server + example Skills for external agents
  • Online demo environment
  • API key desensitization in UI
  • OpenTelemetry OTLP ingestion documentation
  • AVX2 CPU support check at startup
  • Improved AI tool-call success rate
  • Virtual service service.exception display fix
  • Expanded ops docs (upgrade, offline install, Docker/K8s ops)
Loading

v0.1.1

v0.1.1

Choose a tag to compare

View all tags
@databufflabs databufflabs released this 26 Jun 07:04
0.1.1
da0b6fc

Highlights

  • AI multi-agent platform — Ask in natural language; AI Brain dispatches metric, trace, and inspection experts
    • OpenTelemetry-native APM — Services, traces, topology, errors, and dependencies in one UI
    • 5-minute Docker install — Single curl | bash script with amd64/arm64 image bundles
    • Optional demo workload — Sample services (service-a / service-b) for instant traces and topology
    • English UI — Locale switch (en-US) for global users

Quick Start

# Platform
curl -fsSL https://databuff.ai/databuff/ai-apm-install.sh | bash

# Demo apps (optional)
curl -fsSL https://databuff.ai/databuff/ai-apm-demo-install.sh | bash

Open http://YOUR_HOST:27403 · login admin / Databuff@123 · configure LLM API key for AI features.

Kubernetes

curl -fsSL https://databuff.ai/databuff/ai-apm-k8s-install.sh | bash
curl -fsSL https://databuff.ai/databuff/ai-apm-demo-k8s-install.sh | bash

What's included

Component Image
ai-apm-web databuffhub/ai-apm-web:0.1.1
ai-apm-ingest databuffhub/ai-apm-ingest:0.1.1
ai-apm-demo databuffhub/ai-apm-demo:0.1.1

Demo

DataBuff demo

Links

Full changelog

  • Initial public release track 0.1.1
    • Docker / Docker Compose offline install scripts
    • Kubernetes install + demo seeder manifests
    • AI Platform: chat, tools, skills, custom experts
    • APM: service list, global topology, service flow, trace explorer
    • i18n: English + Chinese UI
Loading