Bulk Operations

Tutorial

Bulk Operations

Safely perform bulk delete and management operations

25 min Advanced

AdminBulkDeleteSafety

What You'll Learn

• Bulk Delete — Delete multiple resources with dry-run preview
• Filtering — Select resources by name prefix, status, or creator
• Dry Run vs Execute — Preview matches before committing
• Safety Mechanisms — expected_count guard and 100-resource limit

# Cell 1 — Parameters
USERNAME = "_FILL_ME_IN_"  # Set your email before running

# Cell 2 — Connect
from graph_olap import GraphOLAPClient
client = GraphOLAPClient(username=USERNAME)

# Cell 3 — Provision
from notebook_setup import provision
personas, conn = provision(USERNAME)
analyst = personas["analyst"]
admin = personas["admin"]
ops = personas["ops"]

client = admin
print(f"Connected to: {admin._config.api_url}")

1

Dry Run Preview

Always preview what would be deleted before executing

# Preview what would be deleted (dry run — no resources are removed)
preview = client.admin.bulk_delete(
    resource_type="instance",
    filters={"name_prefix": "test-", "status": "terminated"},
    reason="cleanup old test instances",
    dry_run=True,
)

print(f"Would delete {preview['matched_count']} instances")
print(f"Matched IDs: {preview['matched_ids']}")

2

Filter Types

Select resources by name prefix, status, or creator

# Available filter keys for bulk_delete:
#
#   name_prefix  — match resources whose name starts with a string
#   status       — filter by resource status
#
# Filters are combined with AND logic.

# Example: preview instances matching a prefix
preview = client.admin.bulk_delete(
    resource_type="instance",
    filters={"name_prefix": "cleanup-demo-"},
    reason="cleanup demo instances",
    dry_run=True,
)

print(f"Matched {preview['matched_count']} instances")
if preview["matched_ids"]:
    for iid in preview["matched_ids"]:
        print(f"  - {iid}")
else:
    print("  (none matched — expected for a demo prefix)")

3

Executing the Delete

Move from dry run to actual deletion

# Available filter keys for bulk_delete:
#
#   name_prefix  — match instances whose name starts with a string
#   status       — "terminated", "running", "stopped", etc.
#   created_by   — username of the analyst who created them
#
# Filters are combined with AND logic.

# Example: find terminated instances with a specific prefix
preview = client.admin.bulk_delete(
    resource_type="instance",
    filters={"name_prefix": "cleanup-demo-"},
    reason="cleanup demo instances",
    dry_run=True,
)

print(f"Matched {preview['matched_count']} instances")
if preview["matched_ids"]:
    for iid in preview["matched_ids"]:
        print(f"  - {iid}")
else:
    print("  (none matched — this is expected for a demo prefix)")

4

Safety Mechanisms

expected_count guard and hard limits

# Safety mechanism 1: expected_count
# If the actual matched count differs from expected_count, the API
# returns an error and deletes nothing. This prevents surprises when
# filters match more resources than anticipated.

# Safety mechanism 2: hard limit of 100 resources per call
# The API refuses requests that would delete more than 100 resources
# in a single call. Split larger cleanups into batches.

# Example: expected_count mismatch (would raise an error)
# result = client.admin.bulk_delete(
#     resource_type="instance",
#     filters={"name_prefix": "test-", "status": "terminated"},
#     reason="cleanup",
#     expected_count=5,   # but 12 actually match -> error
#     dry_run=False,
# )

print("Safety mechanisms:")
print("  1. expected_count — must match actual matched_count or delete is refused")
print("  2. Hard limit    — max 100 resources per bulk_delete call")
print("  3. dry_run=True  — always preview before executing")
print("  4. reason        — every delete is audit-logged with a reason string")

Key Takeaways

• bulk_delete() requires resource_type, filters, and reason
• Always run with dry_run=True first to preview matched resources
• Pass expected_count when executing — the API refuses if the count does not match
• A hard limit of 100 resources per call prevents runaway deletions
• Filter by name_prefix, status, or created_by (combined with AND logic)