Hi team, QQ I'm trying to retrieve all of our big...
# troubleshoot
p
Hi team, QQ I'm trying to retrieve all of our bigquery datasets, even those that we didn't directly ingest (i.e they were created from lookml lineage or something) I'm using the below snippet:
Copy code
search(
      input:
          {
              type: DATASET,
              query: "",
              count: 6000,
              filters: [
                  {
                      field: "platform",
                      value: "urn:li:dataPlatform:bigquery"
                  }
              ]
          }
  )
However this is only returning datasets that I have directly ingested, and not those brought in "indirectly" through lineage or other means. Is this supposed to happen? And if so, can I modify the search params to include these results?
b
So I’m guessing you mean the unresolved pointers in the system? We cal these ‘ghost nodes’ - and yes they don’t formally exist as real entities inside DataHub until we have more information (since we can’t render much of use given only an urn)
p
Yeah those! So I can query their URNs directly, but they won't be returned in a search query?
Sorry for ping @big-carpet-38439 but would love resolution on this. We currently depend on a process that uses datahub lineage to determine the impact on dashboards if tables are delayed And looking to see if we can work around this by including "ghost nodes" in this process, but they aren't returning in search results
b
Yeah you can query for those nodes, but the system considers them "virtual" -- that's why you don't get much information back
So why are ghost nodes problematic here? Is it that the upstram table node doesn't exist?
Or that the dashboard nodes are the ghost nodes?
p
The problem is that the ghost nodes are not being returned at all in the above search query
Gentle ping
b
I see
Returning ghost nodes in search results would mean that every unbound edge reference would need to be resolved. However, this is a substantial change to our contract and in many cases may not make sense (e.g. in mistake scenarios where urns are malformed)