def list_nodes():
"""List all nodes."""
configure()
print("")
nodes = classify_nodes()
if nodes[INACTIVE]:
print("")
print(magenta("** INACTIVE nodes **"))
if lb_specified():
print(magenta("INACTIVE nodes are nodes not behind the load balancer."))
elif virtual_ip_specified():
print(magenta("INACTIVE nodes are nodes not behind the virtual IP."))
else:
print(magenta("INACTIVE nodes are nodes in excess of the target cluster size."))
if len(nodes[ORPHAN]) > 0:
print(magenta("To decommission INACTIVE and ORPHAN nodes, use the prune task."))
else:
print(magenta("To decommission these nodes, use the prune task."))
show(nodes[INACTIVE])
if nodes[ORPHAN]:
print("")
print(red("** ORPHAN nodes **"))
if lb_specified():
print(red("ORPHAN nodes are nodes not behind the load balancer but ought to be according to target cluster size and node id sequence."))
elif virtual_ip_specified():
print(red("ORPHAN nodes are nodes not behind the virtual IP but ought to be according to target cluster size and node id sequence."))
if nodes[INACTIVE]:
print(red("To decommission INACTIVE and ORPHAN nodes, use the prune task."))
else:
print(red("To decommission these nodes, use the prune task."))
show(nodes[ORPHAN])
if nodes[EXTRA]:
print(cyan("** EXTRA nodes **"))
print(cyan("EXTRA nodes are nodes in excess of the target cluster size but live behind the load balancer."))
print(cyan("To decommission these nodes, use the scale_down task."))
show(nodes[EXTRA])
if nodes[ACTIVE]:
print("")
print(green("** ACTIVE nodes **"))
if lb_specified():
print(green("ACTIVE nodes are nodes behind the load balancer and current according to target cluster size and node id sequence."))
elif virtual_ip_specified():
print(green("ACTIVE nodes are nodes behind the virtual IP and current according to the node id sequence."))
show(nodes[ACTIVE])
if nodes[ACTIVE] or nodes[ORPHAN] or nodes[EXTRA] or nodes[INACTIVE]:
print("")
print("To decommission all nodes, use the teardown task.")
print("To connect to a node, use:")
print(green("ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -i %s %s@<ip address>" % (env.key_filename[0], env.user)))
else:
print(cyan("There are no nodes in the cluster."))
def classify_nodes():
# Confusion warning: node.id is platform-role-unique_identifier, id_of(node) is unique_identifier.
# For rolling strategy, unique identifers are sequential
cluster_nodes = instances_with_platform_and_role(env.platform, env.role)
cluster_node_ids = set([node.id for node in cluster_nodes])
# max gets grumpy if only 1 argument. Hence two zeros to handle case when cluster is empty.
max_seq_number = max(0, 0, *[id_of(node) for node in cluster_nodes])
if lb_specified():
live_nodes = lb_get_nodes()
elif virtual_ip_specified():
node = virtual_ip_get_node()
live_nodes = [node] if node else []
else:
live_nodes = cluster_nodes
live_node_ids = set([node.id for node in live_nodes])
live_nodes_not_in_cluster = [node for node in live_nodes if not node.id in cluster_node_ids]
# reversed sequential ids means newest to oldest
all_nodes = reversed(sorted(live_nodes_not_in_cluster + cluster_nodes, key = id_of))
active_nodes = []
extra_nodes = []
orphan_nodes = []
inactive_nodes = []
for node in all_nodes:
if node.id in live_node_ids:
if node.id in cluster_node_ids and len(active_nodes) < env.num_nodes:
# ACTIVE:
# live behind load balancer / virtual IP (if applicable)
# in the right platform / role
# still current per desired cluster size and FIFO order
active_nodes.append(node)
else:
# EXTRA:
# live behind the load balancer (thus no extras in Simple or Virtual IP modes)
# ANY platform / role
# NOT current per desired cluster size and FIFO order
if lb_specified():
extra_nodes.append(node)
else:
# INACTIVE:
# NOT behind the load balancer / virtual IP (if applicable)
# in the right platform /role
# NOT current per desired cluster size and FIFO order
inactive_nodes.append(node)
else:
# Node not live. All nodes being processed are in cluster or live, so not live implies in cluster
if len(active_nodes) + len(orphan_nodes) < env.num_nodes:
# ORPHAN:
# NOT behind load balancer / virtual IP (thus no orphans in Simple mode)
# in the right platform / role
# still current per desired cluster size and FIFO order
orphan_nodes.append(node)
else:
inactive_nodes.append(node)
return {
ACTIVE : [node for node in reversed(active_nodes)],
EXTRA : [node for node in reversed(extra_nodes)],
INACTIVE : [node for node in reversed(inactive_nodes)],
ORPHAN : [node for node in reversed(orphan_nodes)],
MAX_ID : max_seq_number
}
