HVR Hub Server Documentation

What is Fivetran HVR?

Fivetran HVR (High-Volume Replicator) is a self-hosted enterprise-level replication platform for databases and file systems. It provides real-time Change Data Capture (CDC) with low latency and high throughput, supporting a distributed architecture across numerous Database Management Systems (DBMSs).

Core Modules

Key Capabilities

• Real-time CDC with sub-second latency
• Supports 50+ database platforms and file systems
• Automatic DDL replication (schema changes)
• Built-in conflict detection and resolution
• Data transformation during replication
• Web UI and REST API for management
• Secure cloud proxy connectivity to Fivetran

Server Specifications

Storage Layout

Network Ports

System Architecture

The HVR Hub Server is the central component that manages all replication channels, schedules jobs, and maintains the configuration repository.

Components on this Server

Hub and Agent Architecture

HVR uses a Hub-Agent model:

• Hub (this server) — Central orchestrator at /usr/hvr/hvr_home. Manages channels, locations, actions, and scheduling.
• Agent (HVA) — A secondary agent installation exists at /usr/hva/hvr_home (HVR 6.2.0/1). Agents run on source/target systems to execute capture and integrate jobs locally.
• Proxy Agent — Java-based agent connecting to Fivetran cloud (wss://prod.aws.us-east-1.proxy.fivetran.com) for managed replication.

Hub Configuration

Directory Structure

OS Users

Systemd Service

[Unit]
Description=HVR Hub Server

[Service]
Type=forking
Environment="HVR_HOME=/usr/hvr/hvr_home"
Environment="HVR_CONFIG=/usr/hvr/hvr_config"
Environment="HVR_TMP=/usr/hvr/hvr_tmp"
User=hvr
ExecStart=/usr/hvr/hvr_home/bin/hvrhubserver
Restart=always
RestartSec=5s

[Install]
WantedBy=multi-user.target

Hub Server Configuration

The hub server configuration is stored in /usr/hvr/hvr_config/etc/hvrhubserver.conf:

{
  "Database_Host": "localhost",
  "Database_Name": "hvr6",
  "Database_Port": 5432,
  "Database_User": "postgres",
  "HTTP_Port": "4340",
  "License_Agreement_Accepted": true,
  "Repository_Class": "postgresql",
  "Repository_Class_Flavor": "aurora_pq"
}

Fivetran Proxy Agent

The proxy agent connects this hub to Fivetran's cloud management plane via a secure WebSocket tunnel:

PostgreSQL Repository

HVR uses a local PostgreSQL 14 instance as its metadata repository. The cockpit displays real-time status including uptime in both Amsterdam (CET/CEST) and Los Angeles (PT) timezones.

NFS Mount

The server has an NFS mount from sapidesecc8 for Oracle archive log access:

# /etc/fstab entry
sapidesecc8:/oracle/ABA/oraarch    /archive    nfs    defaults    0  0

This provides access to SAP ECC Oracle archive logs for CDC replication.

Service Management

Check Status

systemctl status hvrhubserver --no-pager

Start / Stop / Restart

# Start (also re-enable auto-restart)
systemctl enable hvrhubserver && systemctl start hvrhubserver

# Stop (also disable to prevent auto-restart)
systemctl stop hvrhubserver && systemctl disable hvrhubserver

# Restart
systemctl restart hvrhubserver

Process Overview

When running, the following processes should be active:

Web UI Access

HVR provides a web-based user interface for managing replication:

Web UI Sections

REST API

HVR exposes a comprehensive REST API on port 4340 with ~200 endpoints organized into 19 interface categories:

The API uses standard HTTP methods (GET, PUT, POST, DELETE) and returns JSON. It is version-independent and backward-compatible.

Capture

The Capture module continuously reads changes from the source database transaction logs (CDC). Key parameters:

Integrate

The Integrate module applies captured changes to the target location. Supports various target types including databases, Kafka, and file systems.

Actions Reference

Actions define replication behavior. A channel requires at minimum Capture (on source) and Integrate (on target).

Full action reference: fivetran.com/docs/hvr6/action-reference

Key HVR Commands

All commands are in /usr/hvr/hvr_home/bin/. Run as user hvr with environment variables set.

Environment Setup

export HVR_HOME=/usr/hvr/hvr_home
export HVR_CONFIG=/usr/hvr/hvr_config
export HVR_TMP=/usr/hvr/hvr_tmp
export PATH=$HVR_HOME/bin:$PATH

Common Commands

Example: Check Hub Status

# As user hvr
su - hvr
export HVR_HOME=/usr/hvr/hvr_home
export HVR_CONFIG=/usr/hvr/hvr_config
$HVR_HOME/bin/hvrcontrol -h hvrhub list

Overview

HVR uses a local PostgreSQL 14 instance as its metadata repository. This database stores all hub configuration, channel definitions, location settings, replication state, event history, and job metadata.

Service Management

The cockpit provides Start, Stop, and Restart controls for the postgresql-14 systemd service on the hub server.

Manual Service Control (CLI)

# Check status
systemctl status postgresql-14 --no-pager

# Start (enable self-healing)
systemctl enable postgresql-14 && systemctl start postgresql-14

# Stop (disable self-healing)
systemctl stop postgresql-14 && systemctl disable postgresql-14

# Restart
systemctl restart postgresql-14

Backup

Database backups use pg_dump in custom format (-Fc) for efficient compression and selective restore. Each backup is written to two locations simultaneously:

From the Cockpit

Click Backup Now in the PostgreSQL status card. The backup runs in the background and typically completes in a few seconds.

Manual Backup (CLI)

# SSH to the hub server
ssh root@10.128.15.240

# Create backup directories
mkdir -p /usr/hvr/hvr_config/backup /media/HVR
chmod 777 /usr/hvr/hvr_config/backup /media/HVR

# Run pg_dump (custom format for efficient restore)
BKNAME=hvr6_$(date +%Y%m%d_%H%M%S).dump
su - postgres -c '/usr/pgsql-14/bin/pg_dump -Fc hvr6' > /usr/hvr/hvr_config/backup/$BKNAME

# Copy to secondary location
cp /usr/hvr/hvr_config/backup/$BKNAME /media/HVR/$BKNAME
chown hvr:hvr /usr/hvr/hvr_config/backup/$BKNAME /media/HVR/$BKNAME

# Verify both copies
ls -lh /usr/hvr/hvr_config/backup/$BKNAME /media/HVR/$BKNAME

Restore

Restoring from a backup requires stopping the HVR Hub Server first, as active connections prevent a clean restore.

From the Cockpit

Click Restore in the PostgreSQL status card. Select a backup from the list, confirm the operation, and enter the master password. The cockpit automatically stops the hub, restores, re-enables self-healing, and restarts.

Manual Restore (CLI)

# Stop HVR Hub Server
systemctl stop hvrhubserver

# Restore (--clean drops and recreates objects)
su - postgres -c '/usr/pgsql-14/bin/pg_restore --clean --if-exists -d hvr6 /usr/hvr/hvr_config/backup/hvr6_YYYYMMDD_HHMMSS.dump'

# Re-enable and restart
systemctl enable hvrhubserver
systemctl start hvrhubserver

# Verify
systemctl status hvrhubserver --no-pager

Overview

HVR Hub updates are managed through the cockpit's HVR Updates section. The process supports both upgrades and downgrades by uploading a Fivetran HVR release package (.tar.gz) and applying it to the running installation.

Package Format

Release packages are .tar.gz compressed archives containing the full HVR home directory. Each package includes a hvr.ver file with the version string (e.g., Fivetran HVR 6.2.5/5 (linux_glibc2.12-x64-64bit)).

Example filename: fivetran-6.1.0_94-hub_and_agent-linux_glibc2.12-x64-64bit-ga_patch.tar.gz

Update Workflow

The cockpit provides a 4-step workflow for applying updates:

Apply Update — Detailed Steps

When you click Apply Update in the cockpit, the following commands execute on the hub server via SSH:

# 1. Stop the HVR Hub service
systemctl stop hvrhubserver

# 2. Create a timestamped backup of the current installation
cp -a /usr/hvr/hvr_home /usr/hvr/hvr_home.backup.$(date +%Y%m%d_%H%M%S)

# 3. Extract the new package over the existing installation
cd /usr/hvr/hvr_home
tar xzf /usr/hvr/hvr_config/download/<package>.tar.gz

# 4. Fix file ownership
chown -R hvr:hvr /usr/hvr/hvr_home

# 5. Restart the service
systemctl start hvrhubserver

Rollback Procedure

If an update fails or causes issues, restore from the automatic backup:

# Stop the service
systemctl stop hvrhubserver

# List available backups
ls -la /usr/hvr/hvr_home.backup.*

# Restore from the most recent backup
rm -rf /usr/hvr/hvr_home
cp -a /usr/hvr/hvr_home.backup.YYYYMMDD_HHMMSS /usr/hvr/hvr_home

# Restart
systemctl start hvrhubserver

Manual Update (CLI)

If the cockpit is unavailable, you can perform the update manually via SSH:

# SSH to the hub server
ssh root@10.128.15.240

# Upload the package (from your local machine)
scp fivetran-*.tar.gz root@10.128.15.240:/usr/hvr/hvr_config/download/
chown hvr:hvr /usr/hvr/hvr_config/download/*.tar.gz

# Check the staged version
tar xzf /usr/hvr/hvr_config/download/*.tar.gz -O hvr.ver

# Compare with running version
cat /usr/hvr/hvr_home/hvr.ver

# Apply (stop, backup, extract, start)
systemctl stop hvrhubserver
cp -a /usr/hvr/hvr_home /usr/hvr/hvr_home.backup.$(date +%Y%m%d_%H%M%S)
cd /usr/hvr/hvr_home && tar xzf /usr/hvr/hvr_config/download/*.tar.gz
chown -R hvr:hvr /usr/hvr/hvr_home
systemctl start hvrhubserver

# Verify
cat /usr/hvr/hvr_home/hvr.ver
systemctl status hvrhubserver --no-pager

Staging Directory

Version Format

HVR version strings follow the pattern: Fivetran HVR major.minor.patch/build (platform)

The cockpit compares version tuples (major, minor, patch, build) to determine if the staged package is an upgrade, downgrade, or the same version.

Common Issues

Log Locations

External Resources

• Fivetran HVR 6 Documentation — Complete official documentation
• Getting Started Guide — Architecture, installation, first replication
• Action Reference — Complete action and parameter documentation
• REST API Reference — ~200 endpoints for automation
• Supported Platforms — OS and DBMS compatibility matrix
• FAQ & Knowledge Base — Troubleshooting, best practices, how-to guides
• Release Notes — HVR 6.x release history and downloads