Agent stuck in collecting state if data collection is triggered via scheduled job and data collection never comes back to the Instance - Known Error

Agent stuck in collecting state if data collection is triggered via scheduled job and data collection never comes back to the Instance Description Agents can get stuck in "Collecting" state if no payload ever comes back for the collection and the data collection was triggered via a scheduled job, for example: periodic refresh of agent host data, or initial host data collection.

Note: for data collection triggered by the "Collect Host Data" UI action on the agent record, the agent will not get stuck in collecting state. The agent's data collection state will correctly change to "Collection Failed" after the timeout.

For data collections done when ACC-VC is installed, the Enhanced Discovery policy is responsible for refreshing the host data and therefore, agents do not go into "Collecting" state when the data collection is executing, so the scenario does not apply to the automatic data collection when ACC-VC is installed.

Steps to Reproduce

Connect an agent to Instance for the first time. Wait for the initial host data collection to run, the agent will go to "Collecting" state. Turn off both the MID and agent while the data collection is running. Turn on the MID and the agent. Wait the timeout duration of the data collection check. Actual: The agent will be stuck in "Collecting" state.

Expected: The agent should go into "Collection Failed" state after waiting the timeout duration.

This could happen for agents connected to ICS as well, if the data collection payload is lost.

Workaround 1. Use Background script to set any agents stuck in "Collecting" state back to "Collection Failed" state. Then the ACC Admin can trigger data collection again if needed.

2. Apply the attached update set "sys_remote_update_set_1a2f9e5e375722103a56d05174924bae.xml" to the Instance. This update set adds a scheduled job called "ACC Data Collection Timeout Update Set" that runs every hour. The job checks will set the collection status to "Collection Failed" for any agents that are in "Collecting" state for more than an hour.

The update set should be backed out once the customer upgrades to ACC-F version 5.1.0, where we have productized this same Scheduled Job logic.

If the update set is applied, to modify the threshold for marking an agent as "Collection Failed", navigate to the "ACCDataCollectionTimeoutUpdateSet" Script Include and change the value in the addSeconds() function call in the "setAgentsStuckInCollectingToCollectionFailed" function.

// threshold is an hour ageThreshold.addSeconds(-3600); Change the -3600 to whatever the desired threshold is in seconds. Make sure to keep the negative sign. Related Problem: PRB1904373