Skip to content

Discovery Sources

Purpose

Discovery Sources is the Clarity Transform per-source view of discovery: whether each configured integration's runs are healthy, what it contributed to the CMDB, and what hierarchy it found. The on-screen subtitle reads: "Per-source discovery health and what each integration contributed to the CMDB".

The Integrations pages tell you how a source is configured. This page shows what it has produced, one source at a time. It is read-only: nothing on it changes data.

The page is careful to tell a failure apart from an empty result. Where a figure could not be read or measured, it says so ("Not recorded", "Not assessable", or a message that the panel could not be loaded) rather than showing 0.

Where to find it

Discover → Data Analysis → Discovery Sources. The page header is Discovery Sources.

What you see

Info panel

When there is something to report, an info panel at the top gives data-attribution notes. Attribution is how a CMDB record is linked to the integration that discovered it. The notes are:

  • Run records with no technology recorded, which cannot be attributed to any source.
  • "Configured but not in the integration catalogue": technologies that are running discovery under a name the integration catalogue does not list.
  • "Records attributed to an unknown integration": CMDB records that match no configured endpoint and no catalogued integration, so no panel on the page reports them.
  • "Records attributed to a catalogued integration that is not configured here, or is not shown on this page": records from an integration that is known but has no enabled endpoint, or is not shown on this page.

The record notes name up to five integrations with their record counts, then "and N more". If the summary cannot be read, the panel reads "The data-quality summary could not be read. The source list below is unaffected."

Sources

The left panel, Sources ("Configured sources"), lists the configured sources grouped by category, for example CONTAINERS, DATABASE, EXPORT and FILE. Each category heading shows how many sources it holds, and you can select it to show or hide them. A source appears once it has at least one enabled endpoint. If none has, the panel reads "No integrations are configured. A source appears here once it has at least one enabled endpoint."

Each source shows:

  • A status dot for its last run: Succeeded, Warning, Failed, or never run ("Configured but has never run"). Hover over it for the status and the time of the last run.
  • Its name. Where a source groups several technologies, for example a platform's API discovery and its export files, hovering over the name lists them ("Includes: ...").
  • A warning icon if "This endpoint technology has no matching integration catalogue row".
  • The number of configured endpoints.

Selecting a source loads the right-hand panel for it. To give the right-hand panel more room, select Collapse the source list; select Show the source list to bring it back. Your browser remembers this choice.

When the page opens, it selects the source that ran most recently. The selected source is kept in the page address, so you can copy the address to share or bookmark the view of that source. If the address names a feed rather than a source (for example an export-file technology that belongs to a platform), the page opens that platform with the feed highlighted and expanded in Feeds. If the address names a source that is not configured, the page reads " is not configured" and asks you to add an endpoint on the Integrations page. If it names a source that is excluded from this page, because it writes only relationships or is an export, the page reads " is not shown on this page".

Run health

For the selected source:

  • LAST RUN: when the source last completed a discovery run.
  • LAST RESULT: the result recorded for that run. It is free text written by the run, so it can read Success while the run still logged errors.
  • DURATION: how long the most recent run took.
  • TOTAL RUNS: every run ever recorded for the source.
  • SUCCESSES, WARNINGS and ERRORS: the number of successful records, warnings and errors, summed over the recent runs. When the source has more runs than the recent window, the labels add "(last N)", for example "(last 20)". These count records and messages, not runs, so they can be well above N. "Not recorded" means no run recorded a count, which is not the same as 0.
  • STALE RECORDS: the records attributed to the source that the most recent successful run did not return, shown as "N of N assessed". For VMware vSphere and Nutanix AHV inventory nothing removes these records automatically, so this is the signal that the CMDB holds records the platform has dropped. "Not assessable" means there is no successful run to compare against.

Hover over the ? icon beside each label for its definition.

Below the cards is a strip of the recent runs, oldest to newest, coloured by outcome, with the caption "Last N run(s), oldest to newest, of N recorded." Hover over a segment for the run's date, result and duration.

A source that has never run shows "This source has never run" instead of the cards. Its counts are not shown as zero, because nothing has been measured yet.

Feeds

For a source that belongs to a platform, a Feeds section lists each feed: a route by which the platform's data arrives, for example its API or an export file. Each feed shows:

  • A status dot and the feed's name.
  • A summary of its endpoints, for example "3 endpoints, 2 healthy, 1 failed". The summary also counts endpoints "with warnings" and "not yet run or unclear". "No enabled endpoint (disabled or removed)" means the feed is listed only because of past runs.
  • Last run.
  • Its staleness: "N of N records not seen by the last successful run", or "Staleness: not assessable" when there is no successful run to compare against or the feed has several files.
  • "N run(s) not matched to an enabled endpoint (removed, renamed or disabled)", when some runs belong to no current endpoint.

Select a feed's heading to show or hide its endpoints. A feed with more than five endpoints starts hidden. Each endpoint row shows a status dot, its name and address, its last run, its last result ("No result recorded" if there is none) and a strip of its recent runs ("No runs recorded under this name" if there are none).

Sources outside a platform show only the run health cards.

Tabs

  • What was discovered: cards and charts of what the source contributed. They depend on the technology. For Hyper-V, for example, the cards are Datacenters, Clusters, Hosts, Virtual machines, Datastores, Storage used, Networks, Allocated vCPU, vCPU overcommit, VMs with snapshots, Hosts not connected and Records attributed, and the charts are Hosts by hardware model, Hosts by hypervisor version, Datastore capacity: used vs free, Cluster memory: used vs total, Cluster CPU: used vs total and Datastores by file system. Cards and charts with a ? icon explain themselves on hover.
  • Hierarchy: the hierarchy the source discovered, as a tree. It loads the first time you open the tab.

On What was discovered:

  • If some record types could not be counted, a warning reads "N of the CI classes could not be counted (...), so the figures below are incomplete. This is a read failure, not an empty source."
  • If the source has contributed nothing, the tab reads "This source has not contributed any records to the CMDB yet, so there is nothing to chart.", or "Nothing to chart for this source." with a reason for each feed.

When the data comes from two or more management planes (for example several vCenter servers), a filter narrows What was discovered to one of them. Its label names the kind of management plane, for example vCenter. Type in the box to search the list, and select a value. Clear the box to show all of them again; text that does not match a value also clears the filter. The filter applies to What was discovered only, not to run health, Feeds or Hierarchy. It is not shown when there is only one management plane, and it resets when you select another source.

Hierarchy

The tree opens with only its top node: the platform, or the source's name when there are several platform roots. Each node shows its name and, in brackets, the number of records below it. For a platform with several feeds, each root also names the feed its records came through.

  • Select a node to expand or collapse it.
  • Drag to move the tree, and scroll to zoom.
  • Hover over a node for its name, its type, its number of descendants and any details the source recorded.
  • Right-click a datacenter, cluster, host or device to open its details panel.
  • The Restore icon resets the view, and Save as Image downloads the tree as a picture.

Nodes the source could not place in the hierarchy are grouped and coloured in the warning colour. A note above the tree says how many records of each type "could not be placed", and why. A note also appears when more than one platform root is present, because the same platform registered under two endpoint addresses (for example a full name and a short name), or collected by two feeds, appears twice. Roots are never merged automatically.

If no hierarchy view exists for that kind of source yet, the tab says so: this is not a statement that the source discovered nothing. If the tree cannot be loaded, the tab reads "The topology could not be loaded for this source. Reopen the tab to retry."

How to

Check that a source is working

  1. Go to Discover → Data Analysis → Discovery Sources.
  2. In Sources, select the source.
  3. Read Run health first: LAST RUN, LAST RESULT, the ERRORS card and the strip of recent runs. Do not rely on LAST RESULT alone.
  4. Read STALE RECORDS. A high count means the CMDB holds records the platform no longer returns, for example deleted VMs. Review them before planning.
  5. If the source has a Feeds section, check each feed's status, Last run and staleness, and any note about runs not matched to an enabled endpoint.
  6. Open the What was discovered tab to see what the source contributed, then the Hierarchy tab. If the management plane filter is shown, use it to look at one management plane at a time.
  7. Read the info panel's attribution notes, for example for records attributed to a catalogued integration that is not configured.
  8. To change the source's configuration, open its card on Integrations.

Share the view of one source

  1. Select the source in Sources.
  2. Copy the page address from your browser. Anyone who opens it sees the same source.