Administer SQLink

Screenshot note: Screenshots are illustrative. Controls or labels may remain untranslated in the demo release, and pages without a faithful English capture retain the source illustration. Control names and layouts can vary by release.

Manage harvesting, spaces, parsing, preprocessing, edits, and scheduling. Use resource synchronization, parsing tasks, namespaces, manual lineage, and scheduling for detailed procedures.

Implementation sequence

1. Inventory systems, environments, sources, and code.
2. Plan namespaces and default schemas.
3. Harvest metadata and inspect results.
4. Add database, report, or OpenLineage resources to spaces.
5. Create parsing tasks and configure preprocessing and mappings.
6. Resolve report failures and omissions.
7. Add evidence-based relationships beyond automatic coverage.
8. Schedule refreshes and monitor parsing, coverage, and path quality.

Harvesting management

Data sources: configure source, scope, sampling, name enrichment, and blacklist. Check that exclusions do not remove critical dependencies. Inspect task state and sample schemas, tables, and fields after execution.

Reports: test connections before saving, then inspect background execution, counts, and failures.

Files: configure the matching engine, namespace, and schema for DDL or other inputs. For API-submitted files, verify source identifiers and metadata context.

Custom metamodels: confirm types, attributes, and relationships before harvesting. Verify object types and connections afterward to avoid isolated nodes.

Lineage spaces

Inspect saved relationships, database resources, report datasets and connections, and OpenLineage identifiers, inputs, outputs, and events. Resolve missing resources through harvesting rather than long-term manual substitutions.

Lineage tasks

The current English environment did not provide a stable task-form capture for this overview. Use the task labels and settings described above, then confirm the actual form in your deployed release.

Use Script for SQL and scripts, ETL for supported tool exports or task definitions, and Container of User-Defined-Object for saved manual lineage that cannot be parsed again. Record purpose, owner, source, engine, namespace, and execution mode. Test a small sample.

Full import overwrites repeated files and resets them to unparsed, including nested paths. It does not necessarily delete all existing files. Incremental import skips identical content, size, path, and name combinations and updates differences.

Before forced metadata binding, align namespaces, connections, schemas, and naming. Repair context when binding fails rather than selecting objects by name alone.

Preprocessing

Use controlled JavaScript, regular expressions, or models to normalize variables, connections, and irregular text. Reference task variables and connections as required. Compare before and after text, preserve semantics, and retain change records.

Parsing reports

Result	Meaning	Action
Success	Parsing completed	Sample key paths
Partial success	Some statements or objects parsed	Inspect failures, correct, and rerun
Syntax error	Input does not match engine syntax	Check engine, complete code, and variables
Missing metadata	Parsed nodes cannot bind	Harvest missing objects or fix namespace/schema
Other error	Runtime or unknown failure	Inspect logs and retain a reproduction sample

Review parsing success and metadata coverage separately. Map DBLink identifiers to correct connections and namespaces to avoid broken or cross-environment paths.

Namespace and edit management

Record system, environment, resource type, and naming rules. Changing a task's namespace affects matching; assess related resources first. Manual edits require evidence. Current-edit deletion does not remove original canvas content; saved custom files can be viewed or deleted but cannot be edited again.

Scheduling and operations

Select time templates and recurrence. Multiple templates use the intersection of executable dates; deleting or disabling a referenced template removes its constraint. Monitor last and next runs, state, file counts, reports, and repeated failures. Verify an actual trigger after schedule changes.

For new sources, confirm network, permissions, database type, and compatible drivers. Test driver or parser plugin changes in a controlled scope and record affected sources and tasks. Keep passwords out of documentation and screenshots.

Verify results

Check namespace placement, blacklist scope, failure outcomes, separate parsing and coverage acceptance, cross-source mappings, manual edit evidence, and actual scheduled runs.

Troubleshooting

Symptom	Check first
Metadata missing from space	Connectivity, scope, blacklist, task, permissions
File parsing fails	Complete file, engine, variables, preprocessing, logs
Many unmatched nodes	Namespace, schema, harvesting, case
Cross-system breaks	Connections, DBLink, and coverage of both upstream and downstream tasks
Duplicate nodes	Namespace planning, identifiers, import modes
Schedule not triggered	Enabled state, template intersection, dependencies, logs

Related: quick start, exploration, lineage views, scenarios, SQL lineage.