Discovery Identification & re-classification

[now123]

As per our requirement whenever we receive servers from a vendor (DELL/HP/IBM), We have to import them into server class(cmdb_ci_server) with serial number and asset tag, later the build team will install Linux/windows OS into those servers.

If discovery finds the machine it's creating as linux/windows server, still, the cmdb_ci_server table contains the record.   its leads duplicate records.

how the re-classify work in discovery?

[sudharsanv36]

CI Identifiers is what you are looking for. Please check the below link.

Discovery Identifiers - ServiceNow Wiki

[cyked]

Ravi is talking about the identifiers.   Out of the box primary Identification is on Serial Number and Class.   When a server is classified discovery knows if its an AIX server, linux, unix, windows, etc.   When identification happens on a known windows server a lookup to the win server table is done to find a match by serial number.   If a win server is found with that serial number (single match) that CI is updated.   If not match is found at all a new CI is created.   Same for linux OS discovered servers.   If not match by class and SN is found a new CI is created.

When a server is procured its not known what OS will be running on it.   If its brought in through the asset module the model category is like just "server" and a CI will be created on the server table.   When discovery happens identification to "class name" is specific to the OS/class determined.

[VivekSattanatha]

Hi Drew,

I and Ravi worked together on this but till now we didn't find a solution. I agree with your point.  

When a server is procured it's not known what OS will be running on it.   If it's brought in through the asset module the model category is like just "server" and a CI will be created on the server table.   When discovery happens identification to "class name" is specific to the OS/class determined.

This ends up creating duplicate records. One CI in Server class that we created manually and another one in Windows server class while discovering. When discovery runs it should update the existing server class record. instead, it's creating one more record in Windows server table. We are not sure it's a feature or bug.

Regards,

Vivek

[cyked]

I think the issue is more about automating the re-classification of a server.   If you run discovery on a Linux server the gliderecord lookup is to the linux server table, not cmdb_ci_server.   If it did look for ANY server with the returned serial number, i don't think that in and of itself would be a problem.   But if discovery were to try and reclass that server CI to a Linux Server class CI it has implications to the asset record.   Should it try and automate the addition of a model category to the associated product model if Linux Server isn't already one of the categories?

I'm going to set up in Dev using the newer CI Identifiers and see what happens and report back.