outliers: add straightforward remaining columns to node_execution_outliers #81024
Description
Epic: CRDB-13544
Jira issue: CRDB-15392
Via the feature brief, we're looking for crdb_internal.node_execution_outliers to end up looking like the below. Most of these fields will be straightforward to plumb through the outliers system; they are generally already available right at the callsite where we need them. This task is just about those fields; as we encounter difficulty here, let's spin out further issues for the remaining ones.
There is nearby work in #81023, making sure that we honor the user's roles when deciding which of these fields we're willing to show them and which need to be redacted. (Possibly everything is fine except for query?)
Also note that this list hasn't been updated since the design of the v2 insights page. A good next step will be for us to work through that design and see if the columns here still match what we'll need there.
CREATE TABLE node_execution_outliers (
session_id STRING NOT NULL,
transaction_id UUID NOT NULL,
transaction_fingerprint_id STRING NOT NULL,
statement_id STRING NOT NULL,
statement_fingerprint_id STRING NOT NULL,
query STRING NOT NULL,
status STRING NOT NULL,
start_time TIMESTAMP NOT NULL,
end_time TIMESTAMP NOT NULL,
elapsed_time FLOAT8 NOT NULL,
full_scan BOOL,
user STRING NOT NULL,
application_name STRING NOT NULL,
database STRING NOT NULL,
explain_plan JSONB NOT NULL,
rows_read INT8 NOT NULL,
rows_written INT8 NOT NULL,
priority STRING NOT NULL,
contention FLOAT8 NOT NULL,
resource_wait_time FLOAT8 NOT NULL,
retryable BOOL,
retries INT8 NOT NULL,
last_retry_reason STRING,
last_contended_row JSONB,
regions JSONB,
index_recommendation STRING NOT NULL
);