Skip to main content

Issues related to NFS shared resources

The *vds-data/ NFS shared folder is used for sharing the resources within the BEP cluster.

The expected behavior is when BEP Agents start after the BEP Master server and the NFS folder are automatically mounted. However, due to BEP Master server outages, connection abortion, or improper server starting sequence, you can lose the mounting.

So, when you have a problem with the mounting of the NFS shared folder, your BEP Agent disappears from the cluster and becomes not operational for Workers.

NFS folder is not mounted​

Symptoms​

If the NFS folder is not mounted, you get the following error message:

Unable to access jarfile /app/opt/workfusion/vds-data/eval/wfuser/company_name_558e2640-41e2-469e-ae21-a5a900e9018a/lib/vds-spring-boot-app.jar Failed to accept socket: future discarded.

Solution​

To resolve the issue, follow the steps below:

  1. On the BEP Master server, check the shared directory: 

    cd <INSTALL_DIR>/vds-data

  2. Connect to BEP Agent servers and check the NFS directory:

    df -h

  3. Check that the output data contains the following string:

    <BEP_Master_DNS>:<INSTALL_DIR>/vds-data

  4. If the string is missing, it means that the folder is not mounted. To mount a new directory, run the following command:

    mount <INSTALL_DIR>/vds-data/

  5. Check that the mounting is completed successfully:

     df -h

  6. Check that the output data contains the following string:

     <BEP_Master_DNS>:<INSTALL_DIR>/vds-data

    You must also be able to access the sub-directories and files in the new directory.

Slow execution and timeouts​

Symptoms​

You use a non-standard NFS shared folder—external Microsoft Azure or AWS, and you see that it takes too long (for instance, one hour) for the model Workers to start after you launch a Business Process.

Alternatively, you can get the following exception during the model start:

Step name 'Execute AutoML Model: Process Result' has failed. Reason: ' Config line 407: script block java.lang.RuntimeException: Automation extract service unavailable. Http code = 504 responseMap=[body:(http status: 504) Gateway Time-out, status:504]'.

Cause​

It happens due to low connection speed or network throughput limitations. So, the initial downloading of the model resources takes a lot of time.

Solution​

Check the speed and throughput parameters of the network on your side. Contact your system administrator to increase it if needed.

View also: