ACC-V Not Updating Existing CI with Discovery Source = ServiceNow

Aman Trivedi1
Tera Contributor

Hi Community,

We are testing ACC-V on an existing Windows Server CI.


Issue

The ACC agent is Up and Host Data Collection shows Collected, but the existing CI was not being updated.

AmanTrivedi1_3-1793191519342.png

 

The CI had: Discovery Source = ServiceNow

The agent also had only:

Installed Software

and Enhanced Discovery was not assigned.



Finding

The Enhanced Discovery Policy uses: sn_agent.disco_disable_ci_clobber_of_agentless_disco = true

When enabled, the policy filters for:

  • AgentClientCollector
  • ACC-Visibility
  • ServiceNow DEX
    AmanTrivedi1_1-1793191127823.png

     

Therefore, our CI with Discovery Source = ServiceNow was not getting qualified to trigger the Enhanced Discovery check.


Test

For testing only, I manually changed: ServiceNow → ACC-Visibility and triggered Collect Host Data again.

After this:

  • Enhanced Discovery was assigned.
  • Windows CI attributes were updated.
  • MSFT SQL Instance and SQL Server Analysis Services were discovered.
  • Related CMDB relationships were created.

Question

According to the ServiceNow documentation, ACC-V does not perform Discovery when the target already has a CI discovered by Horizontal Discovery with Discovery Source = ServiceNow, unless this behavior is overridden.
Reference: Using Enhanced Discovery and agent-less Discovery together

The documentation also mentions that this setting is ignored if Horizontal Discovery has not run within the configured threshold.

In our case, the last visible Horizontal Discovery update for this CI was more than 14 days before our ACC test, but Enhanced Discovery was still not assigned until we manually changed the Discovery Source to ACC-Visibility.

Existing IRE:

AmanTrivedi1_2-1793191477196.png

 

 

What is the recommended approach here? Should we set this property to false

Thanks!

3 REPLIES 3

kmohammed
Tera Guru

Your diagnosis is correct — this is the clobber protection working as designed, not a failure of the ACC agent. With sn_agent.disco_disable_ci_clobber_of_agentless_disco=true, Enhanced Discovery will not take over a CI whose discovery_source is ServiceNow (agentless Discovery), because ACC-V is not supposed to overwrite agentless data by default. That is why only Installed Software was collected and Enhanced Discovery was never assigned.

 

Do not treat setting the property to false as a blanket fix. It is a source-of-truth decision:

- If ACC-V is now intended to be the authoritative source for these Windows servers, setting the property to false is the supported way to let Enhanced Discovery update agentless CIs. Test it first on a small set of servers and confirm the updated attributes and relationships are what you expect, because agentless values can then be overwritten.

- If agentless Discovery must remain authoritative, leave the property at true. ACC-V will continue to collect host data such as Installed Software without Enhanced Discovery taking over the CI.

 

Also, do not rely on manually changing discovery_source on the CI form as the production approach. That change is only a test shortcut — the next agentless Discovery run can set the source back, and it bypasses the IRE identification rules that govern how the CI should be matched and updated. Make the decision at the policy/property level, then let the next Collect Host Data run apply it consistently.

 

In short: set it to false only where you have deliberately decided ACC-V owns the CI going forward; otherwise keep it true and accept that Enhanced Discovery will not be assigned to agentless CIs.

Thanks @kmohammed 

As we have 400+ servers, what is the best approach to migrate the agent on those prod servers? Any advice

You're welcome, Aman — for 400+ production servers, treat this as a phased migration, not a single push:

 

1. Migrate in waves. Pilot on 10-20 representative servers first (mixed OS versions and locations), validate end to end, then roll out in batches (for example 50-100 per change window). Do not install on the whole estate at once.

 

2. Push the agent with the deployment tool you already use for software distribution (for example SCCM or Ansible), using a silent install that points each agent at the correct instance and MID Server/gateway. Manual installs will not scale and will be inconsistent.

 

3. Control the cutover by wave, not by property alone. The clobber property is global, so scope is controlled by which servers actually have the agent installed. Keep agentless Discovery authoritative during the pilot, and only change the property as part of a planned wave once you have decided ACC-V owns those CIs. Once a wave is validated, pause or retire the agentless Discovery schedule/range for those IPs — otherwise the two sources will keep overwriting each other.

 

4. Validate every wave before starting the next one: agent status = Up, host data collection = Collected, Enhanced Discovery policy assigned, CI attributes updated with discovery_source = ACC-Visibility, and check for duplicate CIs / de-duplication tasks. Also watch MID Server and ECC queue load and stagger collection so 400 agents do not all check in at the same time.

 

5. Keep a rollback for each wave: if validation fails, stop/uninstall the agent for that batch and re-enable the agentless schedule for those IPs. Do not use manual discovery_source edits as the migration method.

 

That way a problem in one wave only affects that batch, and agentless Discovery remains your safety net for the servers you have not migrated yet.