Troubleshoot Replication

Troubleshoot AEM 6.5 LTS Replication

This skill provides systematic troubleshooting guidance for Adobe Experience Manager 6.5 LTS replication issues. Use this to diagnose and resolve problems with content distribution, agent configuration, and replication workflows.

When to Use This Skill

Use this skill when experiencing:

• Replication queues blocked or stuck
• Content not appearing on Publish instances
• Replication agent showing red/error status
• "Replication triggered, but no agent found" errors
• Timeout errors during replication
• Authentication failures (401 Unauthorized)
• Connection refused errors
• Slow or delayed replication
• Dispatcher cache not invalidating
• Reverse replication failures
• Missing or incorrect replication status

Prerequisites

• AEM 6.5 LTS Author and Publish instances
• Administrator access to AEM environments
• Access to replication agent configuration
• Access to log files (error.log, replication.log)
• Understanding of your replication topology

Diagnostic Workflow

Follow this systematic approach to identify and resolve replication issues:

Copy code
1. Verify Symptoms
   ↓
2. Check Agent Status
   ↓
3. Review Replication Queue
   ↓
4. Test Connectivity
   ↓
5. Examine Logs
   ↓
6. Verify Configuration
   ↓
7. Apply Fix
   ↓
8. Validate Resolution

Common Issues and Solutions

Issue 1: Replication Queue Blocked

Symptoms:

• Red status indicator on replication agent
• Queue shows items waiting
• First item in queue failed
• Subsequent items cannot process (FIFO blocking)

Diagnosis:

1. Check agent status:

   Copy code
   Navigate to: Tools → Deployment → Replication → Agents on author
   Look for: Red indicator next to agent name
   

2. View queue details:

   Copy code
   Click agent name
   Review queue entries
   Check error message on failed item
   

Root Causes:

• Network connectivity lost to Publish instance
• Publish instance down or unavailable
• Authentication credentials expired or incorrect
• Insufficient permissions on target content
• Disk space full on Publish
• Large package timeout
• SSL/TLS certificate issues

Solutions:

Solution A: Retry Failed Item

Copy code
Steps:
1. Open blocked replication agent
2. Click "Force Retry" button
3. Monitor queue to see if item processes
4. If successful, remaining items will process automatically

Solution B: Clear Failed Item

Copy code
Steps:
1. Open blocked replication agent
2. Select failed item in queue
3. Click "Clear" to remove it
4. Remaining items will process
5. Manually re-replicate cleared content if needed

Solution C: Restart Replication Components

Copy code
Navigate to: /system/console/bundles
Search for: "replication"

Restart these bundles:
- com.day.cq.cq-replication
- com.day.cq.cq-replication-audit
- com.day.cq.wcm.cq-wcm-replication

Steps:
1. Find bundle
2. Click "Stop"
3. Wait for status: Resolved
4. Click "Start"
5. Verify status: Active

Solution D: Restart Event Processing

Copy code
OSGi Console: /system/console/bundles
Restart: Apache Sling Event Support (org.apache.sling.event)

This clears event queue backlogs

Issue 2: Connection Refused

Symptoms:

• Error: "Connection refused"
• Test connection fails
• Replication queue blocked with connectivity errors

Diagnosis:

1. Verify Publish instance is running:

   bashCopy code
   # Check if Publish is accessible
   curl -I http://publish-host:4503/system/console
   
   # Or browse to:
   http://publish-host:4503/system/console
   

2. Test network connectivity:

   bashCopy code
   # From Author server
   telnet publish-host 4503
   
   # Or
   nc -zv publish-host 4503
   
   # Or
   ping publish-host
   

3. Check replication agent URI:

   Copy code
   Navigate to: Agent → Edit → Transport tab
   Verify: URI matches Publish host and port
   Expected: http://publish-host:4503/bin/receive?sling:authRequestLogin=1
   

Root Causes:

• Publish instance not running
• Firewall blocking connection
• Incorrect hostname or port in agent configuration
• Network routing issues
• DNS resolution failures

Solutions:

Solution A: Start Publish Instance

bashCopy code
cd /path/to/publish/crx-quickstart
./bin/start

Solution B: Fix Network/Firewall

Copy code
1. Verify firewall rules allow Author → Publish on port 4503
2. Check network ACLs and security groups (cloud environments)
3. Verify no proxy blocking connection
4. Test from Author server command line

Solution C: Correct Agent URI

Copy code
Steps:
1. Edit replication agent
2. Transport tab
3. Update URI to correct host/port:
   http://correct-publish-host:4503/bin/receive?sling:authRequestLogin=1
4. Save
5. Test Connection

Issue 3: 401 Unauthorized

Symptoms:

• Error: "401 Unauthorized"
• Authentication failures in logs
• Test connection fails with credential error

Diagnosis:

1. Check agent credentials:

   Copy code
   Agent → Edit → Transport tab
   Verify: User and Password fields
   

2. Verify user exists on Publish:

   Copy code
   Publish instance: http://publish:4503/crx/explorer
   Navigate to: /home/users
   Search for: replication service user
   

3. Check user permissions:

   Copy code
   On Publish instance:
   User → Permissions
   Required: Read, Write, Replicate privileges
   

Root Causes:

• Incorrect username or password
• User doesn't exist on target instance
• User password changed
• User disabled or locked
• Insufficient permissions

Solutions:

Solution A: Update Credentials

Copy code
Steps:
1. Edit replication agent
2. Transport tab
3. Enter correct username
4. Enter correct password
5. Save
6. Test Connection

Solution B: Create/Enable User on Publish

Copy code
On Publish instance:
1. Navigate to: Security → Users
2. Create user: replication-service
3. Set password (match Agent configuration)
4. Save

Grant permissions:
1. Navigate to: Security → Permissions
2. Select user: replication-service
3. Add entries:
   - Path: /content
   - Privileges: jcr:read, crx:replicate, jcr:write
4. Save

Solution C: Reset Password

Copy code
On Publish instance:
1. Navigate to: Security → Users
2. Find user in agent configuration
3. Click "Set Password"
4. Enter new password
5. Save

On Author:
1. Update replication agent with new password
2. Save
3. Test Connection

Issue 4: SSL/TLS Certificate Errors

Symptoms:

• SSL handshake failed
• Certificate validation errors
• HTTPS connection failures

Diagnosis:

1. Check agent URI protocol:

   Copy code
   Agent → Transport tab
   URI: https://... or http://...
   

2. Review error logs:

   Copy code
   error.log contains:
   - javax.net.ssl.SSLHandshakeException
   - PKIX path building failed
   - Certificate validation failed
   

Solutions:

Solution A: Enable Relaxed SSL (Development Only)

Copy code
WARNING: Only for development/testing environments

Steps:
1. Edit replication agent
2. Transport tab
3. SSL section:
   ✓ Relaxed SSL (allow self-signed certificates)
   ✓ Allow expired (allow expired certificates)
4. Save
5. Test Connection

Solution B: Import Certificates (Production)

Copy code
On Author instance:

1. Export certificate from Publish:
   openssl s_client -connect publish:4503 -showcerts > publish-cert.pem

2. Import into Java keystore:
   cd $JAVA_HOME/jre/lib/security
   keytool -import -alias publish-aem -file publish-cert.pem \
     -keystore cacerts -storepass changeit

3. Restart AEM Author

4. Test replication agent connection

Solution C: Use HTTP (Not Recommended for Production)

Copy code
If SSL is not required:
1. Edit agent
2. Transport tab
3. Change URI from https:// to http://
4. Save
5. Test Connection

Issue 5: Content Not Appearing on Publish

Symptoms:

• Replication succeeds (green status)
• Queue processes successfully
• Content still doesn't appear on Publish
• Old content served

Diagnosis:

1. Check content directly on Publish:

   Copy code
   Bypass Dispatcher:
   http://publish:4503/content/mysite/en/page.html
   
   If content appears here but not via Dispatcher:
   → Dispatcher cache issue
   
   If content doesn't appear:
   → Replication issue
   

2. Verify replication status:

   Copy code
   On Author:
   Page → Properties → Basic tab
   Check: Last Published timestamp
   Verify: Status shows "Published"
   

3. Check Publish logs:

   Copy code
   Publish instance: crx-quickstart/logs/error.log
   Search for: path of page
   Look for: Errors during content import
   

Root Causes:

• Dispatcher cache not invalidated
• Dispatcher Flush agent disabled
• Content permissions on Publish
• Sling mapping issues
• Incorrect Publish run mode configuration

Solutions:

Solution A: Manual Dispatcher Cache Clear

bashCopy code
# On Dispatcher server
cd /path/to/dispatcher/cache
rm -rf *

# Or specific path
rm -rf /path/to/dispatcher/cache/content/mysite/en/*

# Check Dispatcher logs
tail -f /path/to/dispatcher/logs/dispatcher.log

Solution B: Verify Dispatcher Flush Agent

Copy code
On Publish instance:
1. Navigate to: Tools → Deployment → Replication
2. Select: Agents on publish
3. Click: Dispatcher Flush
4. Verify: Enabled = ✓
5. Transport tab:
   URI: http://dispatcher:80/dispatcher/invalidate.cache
6. Test Connection
7. If failed, fix connectivity

Solution C: Check Content Permissions on Publish

Copy code
On Publish instance:
1. Navigate to: CRXDE Lite
2. Browse to: /content/mysite/en/page
3. Check node exists
4. Verify permissions: anonymous user can read
5. If not, adjust permissions

Solution D: Force Republish

Copy code
On Author:
1. Select page(s)
2. Manage Publication
3. Action: Unpublish
4. Execute
5. Wait for completion
6. Manage Publication
7. Action: Publish
8. Execute
9. Verify on Publish

Issue 6: Dispatcher Cache Not Invalidating

Symptoms:

• Content published successfully
• Old content served via Dispatcher
• Direct Publish access shows new content
• Dispatcher cache files not deleted

Diagnosis:

1. Check Dispatcher Flush agent:

   Copy code
   Publish instance: /etc/replication/agents.publish/flush
   Status: Should be green (idle/active)
   

2. Review Dispatcher configuration:

   Copy code
   dispatcher.any file:
   /allowedClients {
     /0 { /type "allow" /glob "*publish-ip*" }
   }
   
   /invalidate {
     /0000 { /glob "*" /type "allow" }
   }
   

3. Check Dispatcher logs:

   Copy code
   tail -f /var/log/httpd/dispatcher.log
   
   Look for invalidation requests:
   [date] [I] [pid] Received invalidate request
   

Solutions:

Solution A: Enable Dispatcher Flush Agent

Copy code
On Publish instance:
1. Navigate to: /etc/replication/agents.publish/flush
2. Edit agent
3. Settings tab: ✓ Enabled
4. Serialization Type: Dispatcher Flush
5. Save

Solution B: Fix Dispatcher Configuration

Copy code
Edit dispatcher.any:

/allowedClients {
  /0 { 
    /type "allow" 
    /glob "*<publish-instance-ip>*" 
  }
}

/cache {
  /invalidate {
    /0000 { /glob "*" /type "allow" }
  }
}

Reload Dispatcher:
apachectl graceful

Solution C: Verify Flush Agent Transport

Copy code
Dispatcher Flush agent → Transport tab

Correct URI format:
http://dispatcher-host:80/dispatcher/invalidate.cache

OR if virtual host:
http://www.example.com/dispatcher/invalidate.cache

Test Connection

Issue 7: Timeout Errors

Symptoms:

• Replication timeout errors
• Large packages fail
• Synchronous replication hangs

Diagnosis:

1. Check agent timeouts:

   Copy code
   Agent → Edit → Extended tab
   Connection Timeout: default 10000ms
   Socket Timeout: default 10000ms
   

2. Review package size:

   Copy code
   Large packages (>100MB) may timeout
   Check: crx-quickstart/logs/replication.log
   

Solutions:

Solution A: Increase Timeouts

Copy code
Agent → Edit → Extended tab

Connection Timeout: 30000 (30 seconds)
Socket Timeout: 60000 (60 seconds)

For very large packages: 120000 (2 minutes)

Solution B: Use Asynchronous Replication

Copy code
For large content:
1. Use default async replication (not synchronous)
2. Monitor queue instead of waiting
3. Package-based replication for very large sets

Solution C: Split Large Packages

Copy code
Instead of tree activation:
1. Activate in smaller batches
2. Use incremental replication
3. Schedule large activations during off-peak hours

Issue 8: Replication Triggered But No Agent Found

Symptoms:

• Error: "Replication triggered, but no agent found"
• Content doesn't replicate
• Logs show agent selection failure

Diagnosis:

1. Check enabled agents:

   Copy code
   Navigate to: /etc/replication/agents.author
   Verify: At least one agent is enabled
   Check: Green status indicator
   

2. Review agent triggers:

   Copy code
   Agent → Edit → Triggers tab
   Check: "Ignore default" is NOT checked
   Verify: Appropriate triggers enabled
   

Root Causes:

• All replication agents disabled
• All agents have "Ignore default" checked
• Agent filter excludes all agents
• Custom replication action with invalid agent ID

Solutions:

Solution A: Enable Default Agent

Copy code
Steps:
1. Navigate to: /etc/replication/agents.author/publish
2. Edit agent
3. Settings tab: ✓ Enabled
4. Triggers tab: Uncheck "Ignore default"
5. Save

Solution B: Check Agent Filters (Programmatic)

javaCopy code
// If using ReplicationOptions in code
ReplicationOptions opts = new ReplicationOptions();

// Ensure filter doesn't exclude all agents
opts.setFilter(new AgentFilter() {
    public boolean isIncluded(Agent agent) {
        // Return true for at least one agent
        return !agent.getId().contains("invalid");
    }
});

Solution C: Verify Agent Configuration

Copy code
For each agent:
1. Enabled: ✓
2. Transport URI: Valid and reachable
3. Test Connection: Success
4. Triggers: At least one enabled
5. Ignore default: Unchecked (unless custom workflow)

Issue 9: Replication Not Triggering (Blocked Servlet)

Symptoms:

• Activation button works but nothing happens
• No errors shown but content doesn't replicate
• Replication queues don't populate

Diagnosis:

From official AEM 6.5 LTS documentation: Check for blocking nodes in repository.

1. Check for blocking nodes:

   Copy code
   Navigate to: CRXDE Lite (/crx/de/index.jsp)
   Search for: /bin/replicate or /bin/replicate.json
   
   These nodes may block the replication servlet
   

Root Cause: Custom nodes created at /bin/replicate or /bin/replicate.json can override the default replication servlet, preventing normal replication operations.

Solution:

Copy code
Steps:
1. Navigate to CRXDE Lite: http://localhost:4502/crx/de/index.jsp
2. Check path: /bin/replicate
3. If node exists and is not the system servlet:
   - Right-click node
   - Select "Delete"
   - Save All
4. Repeat for: /bin/replicate.json
5. Test replication

Verification:

Copy code
After deletion:
1. Activate a test page
2. Check replication queue processes
3. Verify content appears on Publish

Issue 10: Namespace Replication Blocked

Symptoms:

• Error: "Namespace replication failed"
• Permission denied on namespace operations
• Custom namespace content won't replicate

Diagnosis:

From official AEM 6.5 LTS documentation: Replication user lacks namespace management privileges.

1. Check user permissions:

   Copy code
   Navigate to: CRXDE Lite
   Path: Repository level (/)
   Check: Replication user privileges
   

Root Cause: The replication user (configured in agent's "Agent User Id") doesn't have jcr:namespaceManagement privilege, which is required to replicate custom namespaces.

Solution:

Copy code
Steps:
1. Navigate to CRXDE Lite
2. Select repository root: /
3. Click "Access Control" tab
4. Find replication service user
5. Add privilege:
   - Privilege: jcr:namespaceManagement
   - Apply
6. Save All

Grant via CRX/DE:

Copy code
1. Tools → Security → Permissions
2. Search for: replication-service user
3. Repository level permissions:
   ✓ jcr:read
   ✓ jcr:write
   ✓ crx:replicate
   ✓ jcr:namespaceManagement  ← Add this
4. Save

Issue 11: Stuck Replication Jobs in Event Queue

Symptoms:

• Multiple agents blocked
• Repository appears to have replication issues
• /var/replication/data has many items

Diagnosis:

From official AEM 6.5 LTS documentation: Check for corrupted replication jobs.

1. Check event queue:

   Copy code
   Navigate to: CRXDE Lite
   XPath Query:
   /jcr:root/var/eventing/jobs//element(*,slingevent:Job)
   
   This shows all pending Sling event jobs
   

2. Check replication data:

   Copy code
   Path: /var/replication/data
   Look for: Large number of nodes
   

Root Cause: Repository corruption or serialization errors can cause replication jobs to get stuck in the Sling event queue.

Solution A: Clean Event Jobs

Copy code
Via CRXDE Lite:
1. Run XPath query:
   /jcr:root/var/eventing/jobs//element(*,slingevent:Job)
   
2. Review results for stuck jobs
3. Identify jobs with:
   - Old timestamps
   - Error properties
   - Replication-related topic
   
4. Carefully delete stuck jobs
5. Save All

Solution B: Clear Replication Data

Copy code
WARNING: Only if queue is irreparably stuck

1. Stop AEM instance
2. Navigate to: crx-quickstart/repository/
3. Backup: /var/replication/data
4. Delete corrupted items in /var/replication/data
5. Start AEM
6. Verify replication resumes

Solution C: Enable Detailed Logging

From official documentation - configure detailed replication logging:

Copy code
Navigate to: /system/console/configMgr
Search for: Apache Sling Logging Logger Configuration

Create new configuration:
- Logger: com.day.cq.replication
- Log Level: DEBUG
- Log File: logs/replication.log

Save and review logs for root cause

Issue 12: Queue Pause Limitations

Symptoms:

• Paused queue resumes automatically
• Pause state lost after restart

Diagnosis:

From official AEM 6.5 LTS documentation: Queue pause has known limitations.

Known Limitations:

1. Not persisted across restarts

   • Pause state is in-memory only
   • AEM restart resumes all paused queues

2. Auto-resume timeout

   • Idle paused queues automatically resume after ~1 hour
   • Not configurable

Workaround:

Instead of pausing, disable the agent:

Copy code
Agent configuration:
1. Edit agent
2. Settings tab
3. Uncheck "Enabled"
4. Save

This persists across restarts

For temporary pause:

Copy code
Accept the limitations:
- Must re-pause after restart
- Must re-pause if idle >1 hour
- Use agent monitoring to track state

Advanced Troubleshooting

Analyze Replication Logs

Location: crx-quickstart/logs/replication.log

Key patterns to search:

bashCopy code
# Successful replication
grep "Replication (ACTIVATE) of /content/mysite" replication.log

# Failed replication
grep "ERROR" replication.log | grep replication

# Agent not found
grep "no agent found" replication.log

# Authentication failures
grep "401" replication.log

# Connection issues
grep "Connection refused" replication.log

Example log analysis:

Copy code
# Find all replication attempts for a path
grep "/content/mysite/en/page" replication.log

# Count failures by type
grep "ERROR" replication.log | cut -d' ' -f5- | sort | uniq -c | sort -rn

# Recent replication activity
tail -100 replication.log | grep "ACTIVATE\|DEACTIVATE"

Monitor via JMX Console

Copy code
Navigate to: /system/console/jmx

Search for: com.day.cq.replication

Monitor MBeans:
- Replication Agent Stats
  - Queue Size
  - Number of queued items
  - Last processed item
  - Error count

- Replication Service
  - Active replications
  - Failed replications
  - Average processing time

Check OSGi Configuration

Copy code
Navigate to: /system/console/configMgr

Relevant configurations:
- Day CQ Replication Service
- Day CQ WCM Replication Impl ReplicationComponentFactoryImpl
- Apache Sling Job Consumer Manager

Verify:
- Services are active
- No configuration errors
- Thread pools not exhausted

Examine Event Queue

Copy code
Navigate to: /system/console/slingevent

Check:
- Event queue depth
- Stuck events
- Processing rate
- Failed events

If queue stuck:
- Restart org.apache.sling.event bundle
- Check disk space
- Review thread dumps

Preventive Measures

1. Regular Agent Testing

Schedule periodic tests:

Copy code
Weekly:
1. Test Connection for all agents
2. Verify queues are empty
3. Review error logs
4. Check disk space on Publish

2. Monitoring and Alerts

Set up monitoring for:

• Replication queue depth > 100
• Agent blocked > 5 minutes
• Replication failures > 10 per hour
• Disk space < 10% free
• High replication lag (>5 minutes)

3. Maintenance Tasks

Regular maintenance:

Copy code
Monthly:
- Review and clear old logs
- Verify agent credentials
- Test disaster recovery procedures
- Update documentation

Quarterly:
- Certificate renewal checks
- Performance testing
- Capacity planning review

4. Best Practices

• Use dedicated service accounts for replication (not admin)
• Implement monitoring for queue depth and errors
• Test agent connectivity after configuration changes
• Document custom agents and their purpose
• Schedule large activations during off-peak hours
• Keep Publish instances in sync with same configuration
• Regular log review to catch issues early
• Maintain runbooks for common issues

Diagnostic Checklist

Use this checklist for systematic troubleshooting:

Copy code
□ Verify symptom and impact
□ Check replication agent status (green/red)
□ Review replication queue for stuck items
□ Test agent connectivity
□ Verify Publish instance is running
□ Check authentication credentials
□ Review error.log and replication.log
□ Verify agent configuration (URI, credentials, settings)
□ Check network connectivity (ping, telnet, curl)
□ Test direct Publish access (bypass Dispatcher)
□ Verify Dispatcher Flush agent (if applicable)
□ Check content permissions on Publish
□ Review OSGi bundles status
□ Examine Sling event queue
□ Check disk space on Author and Publish
□ Verify JVM heap usage
□ Test with simple content first
□ Document findings and resolution

Escalation Path

If issue persists after troubleshooting:

1. Gather diagnostic information:

   • Exact error messages
   • Replication.log excerpts
   • Agent configuration screenshots
   • Steps to reproduce
   • Environment details (AEM version, OS, Java version)

2. Check Adobe Experience League Community:

   • Search for similar issues
   • Post detailed question with diagnostics

3. Adobe Support (if entitled):

   • Open support ticket
   • Provide thread dumps if hanging
   • Share log bundles
   • Include replication queue screenshots

Related Skills

• configure-replication-agent: Set up and configure agents properly
• replicate-content: Understand replication methods
• replication-api: Programmatic replication for custom code

Success Criteria

• ✓ Replication agents showing green status (idle/active)
• ✓ Replication queues empty or processing normally
• ✓ Test Connection succeeds for all agents
• ✓ Content appears on Publish after activation
• ✓ Dispatcher cache invalidates properly
• ✓ No errors in replication.log or error.log
• ✓ Replication status shows "Published" on Author
• ✓ Performance is acceptable (activation <2 minutes)
• ✓ Root cause identified and documented
• ✓ Preventive measures implemented

Additional Resources

• Official AEM 6.5 LTS Replication Documentation
• AEM 6.5 LTS Replication Troubleshooting Guide
• AEM 6.5 LTS Documentation Hub
• Replication API JavaDoc (Package Summary)
• AEM Replication Cookbook (Community)
• Adobe Experience League Community Forums
• AEM Operations Dashboard documentation