Classic UI Siteadmin is slow when replication queue grows on AEM author
Classic UI Siteadmin becomes slow when a publish instance is unreachable and its replication queue grows large on the author instance. As the queue accumulates thousands of entries, requests such as home.pages.json take a long time to complete, which slows down navigation in the content tree and degrades overall authoring experience.
This behavior occurs because Classic UI performs replication‑status inspections as part of UI requests, and large or blocked queues significantly increase processing time. To fix this issue, enable maintenance mode on the affected replication agents so they’re excluded from UI‑level status checks while replication processing continues normally.
Description description
Environment
-
Products
- Adobe Experience Manager (AEM) On‑Premise
- Adobe Experience Manager (AEM) Managed Services (AMS)
-
Versions
- 6.5
- 6.5 LTS
-
Constraints
-
Classic UI Siteadmin is enabled and in use
-
Legacy replication agents exist under:
/etc/replication/agents.author/etc/replication/agents.publish(if publish queues are also managed)
-
Issue / Symptoms
-
Navigating content in Classic UI Siteadmin is slow.
-
In browser Developer Tools, the
home.pages.jsonrequest:- Frequently takes more than 10 seconds to complete when the replication queue is large.
- Gets slower as more items accumulate in the replication queue for the affected publish instance.
-
The replication queue for the affected agent is very long (often thousands of items) because:
- The publish instance is unreachable or down.
- The publish system is undergoing maintenance.
Cause
Classic UI Siteadmin and related status calls perform UI‑oriented inspection of replication queues, including computing queue positions. When a replication queue is very large or blocked, replication status checks become expensive and run on the author request path (for example, while loading home.pages.json). This directly causes slow Classic UI page loads and slow content navigation.
Behavior of maintenanceMode
AEM replication agents support a Boolean flag named maintenanceMode on the agent’s jcr:content node. In AMS, this is surfaced as “Replication Queue Maintenance Mode.”
When maintenanceMode=true on an agent:
- The agent is excluded from replication-status checks in the UI.
- The agent still delivers content; replication processing continues as usual.
- Pending replication for that agent isn’t shown in UI views such as Classic UI Siteadmin, even if the queue is full while others are not.
This mode is designed for scenarios where authors don’t need to monitor a specific agent’s queue, such as when a publish environment is intentionally unreachable during maintenance and a large queue is expected to accumulate.
Resolution resolution
Enable maintenanceMode on the relevant author replication agent(s) whose queues can grow large (typically publish and dispatcher flush agents).
-
Log in to the AEM author instance as an administrator.
-
Open CRXDE Lite:
http(s)://<author-host>:<port>/crx/de -
In the repository tree, navigate to the replication agent configuration, for example:
/etc/replication/agents.author/<agent-name>/jcr:content -
Select the
jcr:contentnode of the affected replication agent. -
Add a new property:
- Name:
maintenanceMode - Type:
Boolean - Value:
true
- Name:
-
Click Save All.
-
Repeat for any additional agents whose queues may grow large during maintenance (for example, failover publish or dispatcher flush agents).
After enabling maintenance mode, verify the following:
home.pages.jsonresponse times in Classic UI Siteadmin are stable and significantly improved.- Authors can navigate the content tree without long delays.
Notes
- No restart is required; changes take effect at runtime.
- While maintenance mode is enabled, detailed queue position and status for that agent aren’t shown in as much detail in the UI, but overall UI responsiveness improves during failover or similar scenarios.