Monitoring Fredhopper – Rezolve Ai Support Portal

This page enables you to setup a full monitoring for Fredhopper in an on-premise installation with the Deployment Package.

Background

Fredhopper can be monitored through log files, via RESTful messages (HTTP), via command line (Deployment Agent) and using the status servlet. This section provides you with a good understanding of the different monitoring access points. Once you understand these possibilities, please follow the instructions below to setup a full monitoring.

Step 1: Before you start

To successfully setup a monitoring system for Fredhopper, the following requirements have to be met:

You have a basic understanding of the possibilities to monitor Fredhopper through log files as well as monitor Fredhopper through HTTP and command line.
You have a basic understanding of the Fredhopper Deployment Package.
Fredhopper is installed successfully with the Deployment Package.
You have decided which monitoring tools to use and know how to setup the monitoring procedures.

Nagios
Fredhopper suggests to use Nagios, see Monitoring FAS using Nagios for more details.
You have a good understanding of your own environment and your reindex processes. The following information should be known in advance:
- Server names / IP addresses.
- Instances names and roles (i.e. etl, deployment, index, live) in Fredhopper.
- Universe names (i.e. catalog01) and locales (i.e. en_GB) in Fredhopper.

Server Names and instance names can be found in the config/topology.txt configuration file. For more information about config/topology.txt see Working with the topology file.

Step 2: Define and setup monitoring

The following sample monitoring setup is based on best practice experience. Please adjust or extend the rules to your specific needs.

Monitor Fredhopper during run-time

This section defines how Fredhopper can be monitored during run-time when no reindex is being processed. Since some processes are stopped during reindex, please pause this monitoring during the time of the reindex.
Example: If the reindex is being processed between 05.00 and 06.00 am, please run the following monitoring rules for 23 hours between 06.00 and 05.00 am.

Group	Schedule	Check	Where	Note
Infrastructure	every 30 sec	if the Deployment Agent is running	all servers	The Deployment Agent should always be active.
Processes	every minute	if the qserver is running	index, live	The qserver needs to run on all index and live instances. This process will be stopped and started during reindex.
		if the xmlloader is running	index	This process will be stopped and started during reindex.
		if the searchindexer is running	index	This process will be stopped and started during reindex.
		if the treebuilder for a universe is running	index	This process will be stopped and started during reindex.
		if the syncserver is running	index	This process will be stopped and started during reindex.
		if the syncclient is running	live	This process will be stopped and started during reindex.
Content	every 5 minutes	if search is available for universe-locale combination	index, live
		if the number of items in instance is correct	index, live
		if the number of items in the searchindex is correct	index, live
		if the number of items in the navigationindex is correct	index, live

Note on incremental updates
When incremental updates are loaded, please note that the number of items might be different on the indexer and the live instances. This is due to small latencies when pushing the data to all servers.

Monitor Fredhopper during reindex

During the reindex, the Fredhopper processes are being stopped and restarted. Therefore, please make sure to apply another set of monitoring rules when a reindex is running.

Schedule	Check	Where	Note
on `bin/run-etl-job`	if the `run-etl-job` command has finished without errors	index	Use process status code (return value/error level) or monitor `data/instances/etl/log/kitchen.log`.
on `bin/reindex`	if the reindex command has finished without errors	index	Use process status code or monitor `data/instances/deployment/log/kitchen.log` and the indexer log files `xmlloader.log`, `treebuilder.UNIVERSENAME.log`, `searchindexer.log`, `qserver.log`.
on `bin/chk-valid-index`	if the generated index satisfies minimum requirements	index	Use process status code
on `bin/fresh-index-to-live`	if the `fresh-index-to-live` command has finished without errors	index, live	Use process status code or monitor `data/instances/deployment/log/kitchen.log` and the indexer log file `syncserver.log` as well as the live instance log files `syncclient.log` and `qserver.log`.

Step 3: Define actions

Each detected, unexpected behavior in Fredhopper should lead to an appropriate action. To finalise your monitoring system, please create an action plan.
Example actions could be defined in the following format:

Detected behavior	Priority	Action
Qserver is inactive on all live instances	1	Send direct message (SMS, E-Mail) to Fredhopper Operations Manager and business contact person with incident report.
Deployment Agent is not running	2	Send direct message (SMS, E-Mail) to Fredhopper Operations Manager with incident report.
Number of items is smaller than expected but higher than minimum.	3	Send email with current and expected number to Fredhopper Operations Manager.