Some Flux customers have IT monitoring suites that support capabilities such as:
If using such a tool to monitor a Flux installation, consider the following guidance for issue identification.
|If you are trying to:||Then use the tool capabilities to:|
|Determine if there are general network connectivity issues|
Ping the server where Flux is executing
Ping the database server
|Determine if Flux engine is down|
Test to see if the Flux service is running by:
|Determine if workflows are failing||Text search in the Flux logs for the term 'threw an exception. The exception is being handled using standard flow chart behavior'. This statement indicates a flowchart has failed and is attempting to recover.|
|Determine if the database server is down||Telnet to the port the database is configured to listen on. If the Telnet connection is successful, it is likely the database is available and operating correctly.|
|Insufficient number of database connections for Flux to properly execute||Text search in the Flux logs for the terms "the engine was unable to get a connection" + "sleeping for X time". This is indicative of there being an insufficient number of database connections being available.|
|Poor performing Flux engine|
Text search in the Flux logs for the term "deadlock". This is indicative of database deadlocks which may be hampering Flux performance.
Confirm that Flux has sufficient memory allocated to it. While a Flux engine will start with 512K, most customers allocate 1 to 1.2 GB to their Flux instances (some significantly more if they are running large or memory-intense Java actions.