Troubleshooting processes#

You can answer a number of questions to help isolate the issue behind problems and process failures.

Take these initial approaches to avoid problems with your processes and check these areas before you take more complicated troubleshooting steps:

• If a process is failing, look for information about the failure in the process request. To open a process request, click the application, and go to the History tab. Find the process in the table, and click View Request. Within the request, expand the steps until you find the step that failed. Failed steps have an error log that provides more information about the failure. All steps have an output log that includes information about the step progress and any logging messages that the step created. For information about viewing logs, see Viewing logs.
• Step through each process element. Verify that you refer to properties by using the correct syntax.
• Verify that the steps are wired together, meaning that each step is connected to another step within a process. Then, verify that the correct step result is selected. That is, on the arrow that connects one step with the next, select the correct condition for proceeding: Run On Success, Run On Failure, or Run on Both.
• Verify that the process is wired in to the Finish step.
• Consider running each process step's equivalent function in a command-line environment to verify that each step works as planned. For example, if the step runs a shell script, try running the shell script on the command line to see whether the same problem happens. If the step starts an application server, make sure that you can start the application server on the command line by using the same commands.
• Verify that you are using the correct component processes type. For information about component process types, see Component process types.
• Trace all the way through your process step-by-step. Steps are essentially commands to be run. See this IBM support article: MustGather: How to trace the execution of a process.
• Verify that no typographical errors are included in the process steps. Review each property and script carefully.
• If properties are a factor in the problem, you can view the properties for an application process by opening the process request and clicking the Properties tab. You can also view the properties that each component process step used by clicking Input/Output Properties . For more information about properties, see Properties and associated topics.
• Check the log for messages that pertain to memory size. You can generate a report of requirements for your operating systems and product version: Hardware requirements for a specific product.
• If you are using sudo commands, permissions can be a problem. Verify that the command is configured correctly. Also, verify that the user is configured in sudo and does not need to specify a password. Make sure that the directory can be accessed by the operating system user that is running the HCL Launch agent, the impersonated user, and the tool that you are calling. For information about sudo commands, see Impersonation on Linux™ systems.
• Review the step output for steps that cause problems. On a command line, run a problem line or step as the same user that HCL Launch uses to run the step or line. Be sure to pass the user and the group.

Other troubleshooting tips#

• If your process branches, look closely at the way the process branches to ensure that the process is running in the way that you expect. See Branching and joining steps.

What issues are you observing or experiencing?#

• A process doesn't start.
• A process fails, but no errors are logged.
• Process failures occur after an upgrade.
• A process runs, but it does not stop.
• Processes fail intermittently.
• Processes stop before completion.
• Processes with branches do not run correctly.
• Processes using impersonation do not run correctly.

A process doesn't start

• Can you run the process while you use the built-in ID? Running the process that way checks for a permission issue on the resource.
• Is the agent running? For information about agents, see Agents.
• Can any processes run on the agent? If required, create a simple “echo Hello World” shell step, and see whether you can run it on the agent. Running this shell step confirms that communication between the server and resource are established.
  • If the simple shell step works, what's different between the working step and the failing step?
• Are the components mapped? See Mapping resources and components to environments.
• Are any errors logged in the step output? See Viewing logs.
• Check the server log file for errors, permissions issues, or configuration issues. See Viewing logs.

A process fails, but no errors are logged

• Confirm that the version of HCL Launch and the agents and agent relays are the same.
• Check the agent log file for errors that point to a configuration issue. For example, a port might be blocked or a URL is incorrect. Check the server log file for errors. For example, a port might be blocked or URL is incorrect. See Viewing logs. See also Firewall and communication configuration.
• If you are using high-availability, check to see whether there was a failover.
• If this result happens for only this process, what might be different between this process and the other processes that succeed at running the step?
• Can the process succeed when the built-in admin ID runs it. When you run the process this way, incorrect permission configuration does not impact process execution.
• Does this problem happen with any other steps? If not, how is this step different?
• If you run a process with just this step, does it fail in the same way?

Process failures occur after an upgrade

• Does the new version support the installed Java™ Runtime Environment or Java Development kit?
• Were there any errors in the upgrade process?
• Check the agent log file for errors that point to a configuration issue. For example, a port might be blocked or a URL is incorrect. Check the server log file for errors. For example, a port might be blocked or URL is incorrect. See Viewing logs.
• If you use an agent relay, confirm that the agent relay was upgraded to the same version as the server.
  • Check the Agent Relays tab to verify that the agent relay version is supported. To open the Agent Relays tab, click the main Resources tab, and then click the Agent Relays tab. See the minimum recommended and minimum required versions for the listed agent relays in the upper-right corner of the tab. You might have to upgrade an agent relay or agent relays. For information about upgrading an agent relay, see Upgrading agent relays from the command line.
• Check the Agents tab to verify that the server supports the agent version. To open the Agents tab, click the main Resources tab, and then click the Agents tab. See the minimum recommended and minimum required versions for the listed agents in the upper-right corner of the tab. You might have to upgrade an agent or agents. For information about upgrading an agent, see Upgrading agents.
• Did you create this process in the upgraded version? If the process was created in a previous version, did it work correctly in the previous version? If so, were any changes made to the process after the upgrade?

A process runs, but it does not stop

• Has this process ever run to completion previously? See this IBM support article: MustGather: How to trace the execution of a process.
• Is the process wired to the Finish step?
• If the step that fails is a shell step, is the standard output is redirected for child processes to https://developer.ibm.com/urbancode/plugindoc/ibmucd/shell/5-562693/usage/?

Processes fail intermittently

• Does the server log contain exceptions that this issue generated?
• On which step does the process fail? Does the step output contain errors?
• At which times does the process fail? Check for network outages, maintenance intervals during that time.
• Is the Download Artifacts step failing with the error java.io.FileNotFoundException: Artifact set not found? Ensure that CodeStation has enough time to store the artifacts before they are requested, particularly in high-availability clusters, in which the servers share the artifacts. The solution for this issue is typically to add a Wait step with a few seconds wait time before the Download Artifacts step.

Processes stop before completion

• Check the agent and server log files for errors.
• Check the process system output.
• Consider running each process step's equivalent function in a command-line environment to verify that each step works as planned.
• Can the built-in admin ID run processes with success?
• If the steps that fail are shell steps, redirect the standard output of child processes to https://developer.ibm.com/urbancode/plugindoc/ibmucd/shell/5-562693/usage/.

• Do you have sufficient memory on the HCL Launch server?

  Tip: The workflow execution engine that runs the process runs on the HCL Launch server.

• On which screen or activity do you see no progress?

• How long did the process stop? How long do you expect this activity to take?
• Are there any agents for which this behavior does not happen? If so, what differences exist between the processes that work and the processes that do not.
• Did this behavior always happen? If not, what changed immediately before this problem surfaced?

Processes with branches do not run correctly

• Look closely at the way the process branches to ensure that the process is running in the way as expected. See Branching and joining steps. See this IBM support article: MustGather: How to trace the execution of a process.

Processes using impersonation do not run correctly

• Reproduce the problem with a simple shell step, use the #! syntax at the top to specify the shell to use. For example:

  ```

  !/bin/sh#

  ```

  In the example, /bin/sh resolves to different shells depending on the operating system. For example, it resolves to /bin/ksh on AIX and Solaris operating systems and to /bin/bash on a RedHat Enterprise Linux operating system.

• Put the shell script of the step into a separate file and execute it from the HCL Launch interface or from a command line.

• Enable logging on the agent to verify that the command ran. To enable logging put the following line in the agent-home/conf/agent/log4j.properties file. Start the agent again for the change to take affect.

  log4j.logger.com.urbancode.shell.impersonation.unix.UnixImpersonateCommand=DEBUG

• Add to the separate script file a call to printenv and execute the script from a HCL Launch step or from a command line. Compare the two processes for differences.

• Execute the separate script using the commands configured in the sudoFormat and suFormat agent properties described in the Impersonation on Linux and UNIX systems topic.

• In the agent installed.properties file, add the following property to verify that the command executes successfully after the impersonation.

  com.urbancode.shell.impersonation.unix.scriptPostExecuteAction=echo

• Update your Java fixpack level or Java version to a higher level.

Parent topic: Troubleshooting the HCL Launch server and agents