fabric-slice-manage

name	fabric-slice-manage
description	List, inspect, renew, modify, and delete FABRIC experiment slices
allowed-tools	["Read","Grep","Glob","Write","Edit","Bash"]

Instructions

When invoked, generate code for FABRIC slice lifecycle management operations. Determine which operation the user needs:

• List: Show all slices or filter by state
• Inspect: Get details of a specific slice (nodes, networks, interfaces)
• Renew: Extend a slice lease
• Modify: Add or remove nodes/networks from a running slice
• Delete: Remove a slice or all slices

Always assume FABlib is initialized as fablib.

API Reference

Listing Slices

fablib.list_slices(
    excludes: list[SliceState] = None,  # States to exclude
    output: str = None,                  # "text", "pandas", "json", "list"
    fields: list[str] = None,           # Fields to display
    quiet: bool = False,
    filter_function=None,
    pretty_names: bool = True,
)

Getting a Slice

# By name
slice = fablib.get_slice(name="my-slice")

# By ID
slice = fablib.get_slice(slice_id="abc-123-def")

# Get multiple slices
slices = fablib.get_slices(
    excludes: list[SliceState] = None,
    slice_name: str = None,
    slice_id: str = None,
    user_only: bool = True,
)

Slice Information

slice.show(fields=None, output=None, quiet=False, colors=False, pretty_names=True)
slice.list_nodes(output=None, fields=None, colors=False, quiet=False, filter_function=None, pretty_names=True)
slice.list_networks(output=None, fields=None, colors=False, quiet=False, filter_function=None, pretty_names=True)
slice.list_interfaces(output=None, fields=None, quiet=False, filter_function=None, pretty_names=True, refresh=False)

Accessing Nodes

node = slice.get_node(name="node1")
nodes = slice.get_nodes()

# Execute commands on a node
stdout, stderr = node.execute("hostname")

# SSH command for manual access
print(node.get_ssh_command())

# Management IP (IPv6)
print(node.get_management_ip())

Renewing a Slice

# Renew by number of days
slice.renew(days=14)

# Renew to a specific end date (UTC)
from datetime import datetime, timezone, timedelta
end_date = (datetime.now(timezone.utc) + timedelta(days=6)).strftime("%Y-%m-%d %H:%M:%S %z")
slice.renew(end_date=end_date)

Modifying a Slice

# Add a new node to running slice
new_node = slice.add_node(name="node3", site="TACC", cores=4, ram=16, disk=50)
slice.modify(
    wait=True,
    wait_timeout=600,
    wait_interval=10,
    progress=True,
    post_boot_config=True,
)

# Accept modifications (required after modify)
slice.modify_accept()

Deleting Slices

# Delete a specific slice
slice.delete()

# Delete all slices (with confirmation progress)
fablib.delete_all(progress=True)

Slice States

from fabrictestbed.slice_manager import SliceState

# Common states:
# SliceState.StableOK - Running normally
# SliceState.StableError - Error state
# SliceState.Configuring - Being set up
# SliceState.Nascent - Just created, not submitted
# SliceState.Dead - Deleted/expired

SSH-Related Methods

# Wait for all nodes to be SSH-accessible
slice.wait_ssh(timeout=1800, interval=20, progress=False)

# Test SSH on all nodes
slice.test_ssh() -> bool

# Run post-boot config (SSH keys, etc.) — only after non-blocking submit
slice.post_boot_config()

# Update slice state from server
slice.update()

Patterns

List Active Slices

fablib.list_slices()

List Only Running Slices

from fabrictestbed.slice_manager import SliceState

fablib.list_slices(
    excludes=[SliceState.Dead, SliceState.Closing],
    output="text",
)

Filter Slices by State

fablib.list_slices(
    filter_function=lambda x: x["state"] == "StableOK"
)

List as Python Objects

slice_list = fablib.list_slices(output="list", quiet=True)
for s in slice_list:
    print(f"Slice: {s['id']}, {s['name']}, {s['state']}")

Inspect a Slice

slice = fablib.get_slice(name="my-experiment")
slice.show()

print("\n--- Nodes ---")
slice.list_nodes()

print("\n--- Networks ---")
slice.list_networks()

print("\n--- Interfaces ---")
slice.list_interfaces()

Renew a Slice

from datetime import datetime, timezone, timedelta

slice = fablib.get_slice(name="my-experiment")

# Extend by 14 days
slice.renew(days=14)

# Verify new end date
print(f"Lease End: {slice.get_lease_end()}")

Execute Commands on All Nodes

slice = fablib.get_slice(name="my-experiment")

for node in slice.get_nodes():
    stdout, stderr = node.execute("uname -a")
    print(f"{node.get_name()}: {stdout}")

Upload/Download Files

node = slice.get_node(name="node1")

# Upload
node.upload_file("local_script.sh", "/home/ubuntu/script.sh")
node.execute("chmod +x /home/ubuntu/script.sh && /home/ubuntu/script.sh")

# Download
node.download_file("local_results.tar.gz", "/home/ubuntu/results.tar.gz")

Add Node to Existing Slice

slice = fablib.get_slice(name="my-experiment")

# Add a new node
new_node = slice.add_node(name="worker3", site="TACC", cores=8, ram=32, disk=100)

# Submit modification
slice.modify()

# Accept the modification
slice.modify_accept()

# Verify
slice.list_nodes()

Delete a Slice

slice = fablib.get_slice(name="my-experiment")
slice.delete()
print("Slice deleted successfully")

Delete All Slices

# WARNING: This deletes ALL your slices
fablib.delete_all(progress=True)

Reconnect After Reboot

# After a node reboot, wait for SSH to come back
slice.wait_ssh(timeout=360, interval=10, progress=True)
slice.update()
slice.test_ssh()