Unknown Failure but execution succeeds

Document created by mike_aronson Employee on Aug 6, 2014
Version 1Show Document
  • View in full screen mode
In Manage -> Process Reporting, a process execution gives an aborted error or an Unknown Failure in a molecule or atom cloud, but then the execution status changes and it shows as Successful at a later time.

This behavior has been observed in the past and may be due to the following.  The molecule or clouds nodes periodically write to files on the file share hosting the molecule or cloud installation folder.  This folder may not always be making the file available to the atom nodes. The error is often occurring in a basic java io method that the boomi atom code calls. For example, see below:
MON DD, YYYY HH:mm:SS AM/PM TIMEZONE SEVERE [com.boomi.execution.ExecutionManager handleFailedExecution] Aborted process ExecutionSubmission: ExecutionTaskConfig AccountID, proc: id, id: execution-id(dir: processes/component-id, source: SCHEDULE), ExecutionDetails Process Name, Node ID, Account[Account ID: Unknown failure occurred
com.boomi.execution.ExecutionException: Unknown failure occurred
MON DD, YYYY HH:mm:SS AM/PM TIMEZONE SEVERE [com.boomi.execution.forker.MultiProcessExecutionFuture checkForResponse] Failed reading execution response file
java.io.FileNotFoundException: <atom installation folder>/execution/history/YYYY.MM.DD/execution-id/executionResponse.dat (No such file or directory)
at java.io.FileInputStream.open(Native Method)
The most common reasons for this type of issue are:
1. intermittent network issues between the servers hosting the node(s) and the file server
2. I/O on the file system is not keeping up with the application requests. this could be because of the hardware or file system is not sufficiently meeting the needs or I/O demand on the file server not related to the atom could be causing
performance issues.
3. An Anti-virus program, file backup program, or similar conflicting program is running on the file server and causing intermittent issues with accessing files.
The process executions eventually show as completed in process reporting, which indicates that the files on fileshare that the molecule or cloud node(s) need to read and write from the execution/history folder become accessible again, after the external issues not related to the atom molecule or cloud node resolve on their own.

Some options that may help with this include enabling local storage as described at these links below. This may reduce the I/O traffic to/from the fileshare:



Also, you may want to enable JMX monitoring on the clouds and monitor system vitals to see the load occurring on the system.You may need to consider adding nodes or clouds to spread execution load across other clouds if you find that there is too much processing on the cloud.

Here is a link that describes how to enable JMX Monitoring:
1 person found this helpful