Triggering Index Replication

Use

By default index replication is deactivated. You can trigger replication in various ways. The table below gives an overview of the methods and of when you should use them.

	Procedure	Effect	Use with
			Indexing with QS*	Indexing without QS*
1.	Manual replication	Index changes are available for searching when you have replicated the index. You use this method for the initial replication.
2.	Automatic replication following optimization in the queue server ( Replicate after Optimize).	Following the optimization of documents, index changes are available in the queue server for the search.
3.	Automatic replication following every index update	All index changes are available quickly for the search.
4.	Time-dependent replication using the queue server	Index changes are available for searching when the next replication has taken place. Replication takes place regularly according to a defined schedule.
5.	Time-dependent replication using the Python scheduler	Index changes are available for searching when the next replication has taken place. Replication takes place regularly according to a defined schedule.
6.	Replication triggering by the application using TREX	TREX provides the (ABAP/Java) application with methods for triggering replication.

*QS = queue server

The system replicates the entire index for the initial replication. In subsequent replications the system only replicates the changed index files. The duration of the replication and the generated system load depends on the following factors:

Are the indexes stored centrally or are they distributed?

With decentralized data storage the replication generates a higher net load because the system has to copy the indexes to the slave hosts.
How often is the index updated?
How many index files need to be replicated? This depends on the size of the index or delta index.
How many indexes need to be replicated?
How large are the indexes? What type of documents are indexed (documents with attributes only, documents with attributes and text content, or only documents with text content). Does the index contain text-mining information?
How quickly should the updated information be available for searching?

In order to determine the optimum time for replication, you have to weigh up the required topicality against the system load generated.

We recommend that you carry out the initial replication manually, since it can last a lot longer than subsequent replications.

If large indexes need to be replicated frequently, it may not be possible for the system to keep to your configured interval for replication. If this is the case, the system carries out automatic replication at the next possible point in time.

Procedure

1. Replicate an Index Manually

Go to the Index Landscape window in the TREX admin tool.
Work through one of the following steps:
- To replicate all indexes, choose Replicate All.
- To replicate a single index, select the index in question and choose Replicate Index from its context menu.

2. Replicate the Index Automatically Following Optimization in the Queue Server ( Replicate after Optimize).

Replicate after Optimize

Set the Replicate After Optimize queue parameter to On.

For more information on changing queue parameters, see Configuring Queue Parameters.

We recommend that you arrange for the index to be replicated automatically after every update - as described in point 3 - rather than using the Replicate After Optimize procedure. Replicate After Optimize only replicates changes involving the TREX queue server. Changes made without involving the queue server, such as changes to index properties and taxonomies, are not replicated.

3. Replicate an Index Automatically Immediately After Every Update

When you create an index, you can arrange for it to be replicated automatically immediately after every update. To do so, proceed as follows.

Go to Index area in the Landscape Configuration window of the TREX admin tool.
Activate automatic index replication by selecting the Auto Replication checkbox.

You can change this setting later on if you want.

Go to the Index Landscape window in the TREX admin tool.
Use the secondary mouse button to click on the index whose index replication settings you want to change.
Choose Landscape Configuration and then Enable Auto Replication or Disable Auto Replication.

This way of triggering index replication is particularly important in scenarios that do not use a TREX queue server.

4. Replicate the Delta Index Time-Dependently

Enter the time at which the replication is to take place in the Replication Time queue parameter.

Use All-3 to trigger replication every three hours. Use All (03:00:00 AM) to trigger replication every morning at 03:00:00 AM.

For more information on changing queue parameters, see Configuring Queue Parameters.

5. Use the Python Scheduler to Schedule Index Replication

Change the following configuration files on all master name servers:

Configuration File

Change

TREXDaemon.ini

If the Python scheduler is not yet active, activate it now:

[daemon]

programs=<other_sections> ,cron

crontab.ini

Remove the comment sign from the following line:

<schedule> python replicate.py silent allIndexes=1 ''

The default setting causes the system to check for changes to an index every 5 minutes. If there are no changes, the system takes no further action. If changes have taken place, they are replicated.

Modify the schedule if necessary. For information on syntax and for examples, see the configuration file.

Result

You can monitor index replication in the TREX admin tool (standalone) in the Index Landscape window. If necessary, you can terminate replications in progress there.