How to fix WebI task slowness when copying the previous snapshot (multi-threading / Oracle)

How to fix WebI task slowness when copying the previous snapshot (multi-threading / Oracle)

Symptoms

The 360Eyes tasks are blocked at the same step in the execution log (for example, they stop after CHOOSE_LOADING_MODE starts).

How to replicate the issue

  1. Run a  Webi 360Eyes task that includes snapshot processing/copy.

  2. Monitor the task execution log.

  3. Observe that multiple tasks remain stuck around the same step, for example:

    • [08/11/2024 20:18:58] Start of COMPUTE_SNAPSHOT_ID step

    • [08/11/2024 20:18:58] End of COMPUTE_SNAPSHOT_ID step

    • [08/11/2024 20:18:58] Start of CHOOSE_LOADING_MODE step

  4. Check the WebI log and notice the snapshot copy phase is extremely slow (can take many hours, up to a day).

Cause

During the snapshot copy, the default multi-threaded processing can cause performance issues with Oracle, leading to very long snapshot copy times and tasks appearing blocked at the same step.

Resolution

Option 1: Reduce snapshot copy threads to 1 (all versions)

  1. Go to the Overridden Parameters section for the webi task/job

  2. Update the parameter to force single-thread processing during snapshot copy:

    • -threads=1

  3. Save the job configuration.

  4. Re-run the 360Eyes task and confirm that the steps progress normally.

Option 2: Use the dedicated parameter (360Suite 2025.2 and higher)

  1. Go to the Overridden parameters section for the webi task/job.

  2. Add the parameter:

    • -nbThreadsWhenCopyingSnapshot=1

  3. Save the configuration.

  4. Re-run the task.

This option is especially relevant for Oracle environments.

Additional information

If tasks still take unusually long after setting threads to 1, and -nbThreadsWhenCopyingSnapshot=1,re-check the WebI logs around the snapshot copy step to confirm the parameter is applied and that the slowness is not coming from storage, network latency, or database load.