evio/controller/modules/GraphBuilder.py

230 lines
10 KiB
Python

# EdgeVPNio
# Copyright 2020, University of Florida
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
import math
import random
from .NetworkGraph import ConnectionEdge
from .NetworkGraph import ConnEdgeAdjacenctList
class GraphBuilder():
"""
Creates the adjacency list of connections edges from this node that are necessary to
maintain the Topology
"""
def __init__(self, cfg, top=None):
self.overlay_id = cfg["OverlayId"]
self._node_id = cfg["NodeId"]
self._peers = None
# enforced is a list of peer ids that should always have a direct edge
self._enforced_edges = cfg.get("EnforcedEdges", [])
# only create edges from the enforced list
self._manual_topo = cfg.get("ManualTopology", False)
self._max_successors = int(cfg["MaxSuccessors"])
# the number of symphony edges that shoulb be maintained
self._max_ldl_cnt = int(cfg["MaxLongDistEdges"])
self._max_ond = int(cfg["MaxOnDemandEdges"])
# Currently active adjacency list, needed to minimize changes in chord selection
self._nodes = []
self._my_idx = 0
self._top = top
self._relink = False
if self._manual_topo and not self._enforced_edges:
self._top.log("LOG_WARNING", "Ad hoc topology specified but no peers are"
"provided, config=%s", str(cfg))
def _build_enforced(self, adj_list):
for peer_id in self._enforced_edges:
ce = ConnectionEdge(peer_id, edge_type="CETypeEnforced")
adj_list.add_conn_edge(ce)
def _get_successors(self):
""" Generate a list of successor UIDs from the list of peers """
successors = []
num_peers = len(self._peers)
num_nodes = len(self._nodes)
successor_index = self._my_idx + 1
num_succ = self._max_successors if (num_peers >= self._max_successors) else num_peers
for _ in range(num_succ):
successor_index %= num_nodes
successors.append(self._nodes[successor_index])
successor_index += 1
return successors
def _build_successors(self, adj_list, transition_adj_list):
num_ideal_conn_succ = 0
successors = self._get_successors()
suc_ces = transition_adj_list.filter([("CETypeSuccessor", "CEStateConnected")])
# add the ideal successors to the new adj list
for peer_id in successors:
if peer_id not in adj_list:
adj_list[peer_id] = ConnectionEdge(peer_id, edge_type="CETypeSuccessor")
if peer_id in suc_ces:
# this is an ideal succ that was previously connected
num_ideal_conn_succ += 1
del suc_ces[peer_id]
# do not remove the existing successor until the new one is connected
for peer_id in suc_ces:
# these are to be replaced when the ideal ones are in connected state
if num_ideal_conn_succ < self._max_successors:
# not an ideal successor but keep until better succ is connected
adj_list[peer_id] = ConnectionEdge(peer_id, edge_type="CETypeSuccessor")
num_ideal_conn_succ += 1
else:
break # consider selecting the best of these
@staticmethod
def symphony_prob_distribution(network_sz, samples):
"""exp (log(n) * (rand() - 1.0))"""
results = [None]*(samples)
for i in range(0, samples):
rnd_val = random.random()
results[i] = math.exp(math.log10(network_sz) * (rnd_val - 1.0))
return results
def _get_long_dist_links(self, num_ldl):
# Calculates long distance link candidates.
long_dist_links = []
net_sz = len(self._nodes)
if net_sz > 1:
num_ldl = min(num_ldl, net_sz)
node_off = GraphBuilder.symphony_prob_distribution(net_sz, num_ldl)
for i in node_off:
idx = math.floor(net_sz*i)
ldl_idx = (self._my_idx + idx) % net_sz
long_dist_links.append(self._nodes[ldl_idx])
return long_dist_links
def _build_long_dist_links(self, adj_list, transition_adj_list):
# Preserve existing incoming ldl
# handled in net builder
#ldlnks = transition_adj_list.edges_bytype(["CETypeILongDistance"])
#for peer_id, ce in ldlnks.items():
# if ce.edge_state in ("CEStateInitialized", "CEStateCreated", "CEStateConnected") and \
# peer_id not in adj_list:
# adj_list[peer_id] = ConnectionEdge(peer_id, ce.edge_id, ce.edge_type)
# evaluate existing ldl
ldlnks = {}
if not self._relink:
ldlnks = transition_adj_list.edges_bytype(["CETypeLongDistance"])
num_existing_ldl = 0
for peer_id, ce in ldlnks.items():
if ce.edge_state in ["CEStateConnected"] and \
peer_id not in adj_list and not self.is_too_close(ce.peer_id):
adj_list[peer_id] = ConnectionEdge(peer_id, ce.edge_id, ce.edge_type)
num_existing_ldl += 1
if num_existing_ldl >= self._max_ldl_cnt:
return
num_ldl = self._max_ldl_cnt - num_existing_ldl
ldl = self._get_long_dist_links(num_ldl)
for peer_id in ldl:
if peer_id not in adj_list:
ce = ConnectionEdge(peer_id, edge_type="CETypeLongDistance")
adj_list.add_conn_edge(ce)
def _build_ondemand_links(self, adj_list, transition_adj_list, request_list):
ond = {}
# add existing on demand links
existing = transition_adj_list.edges_bytype(["CETypeOnDemand"])
for peer_id, ce in existing.items():
if ce.edge_state in ("CEStateInitialized", "CEStatePreAuth", "CEStateAuthorized", \
"CEStateCreated", "CEStateConnected") and peer_id not in adj_list:
ond[peer_id] = ConnectionEdge(peer_id, ce.edge_id, ce.edge_type)
task_rmv = []
for task in request_list:
peer_id = task["PeerId"]
op = task["Operation"]
if op == "ADD":
task_rmv.append(task)
if peer_id in self._peers and (peer_id not in adj_list or
peer_id not in transition_adj_list):
ce = ConnectionEdge(peer_id, edge_type="CETypeOnDemand")
ond[peer_id] = ce
elif op == "REMOVE":
self._top.log("LOG_DEBUG", "Processing OND Removal, popping %s", peer_id)
ond.pop(peer_id, None)
if peer_id not in adj_list:
# only clear the task after the tunnel has been removed by NetworkBuilder
task_rmv.append(task)
for peer_id in ond:
if peer_id not in adj_list:
adj_list[peer_id] = ond[peer_id]
for task in task_rmv:
request_list.remove(task)
def build_adj_list(self, peers, transition_adj_list, request_list=None, relink=False):
self._relink = relink
self._prep(peers)
if request_list is None:
request_list = []
adj_list = ConnEdgeAdjacenctList(self.overlay_id, self._node_id,
self._max_successors, self._max_ldl_cnt, self._max_ond)
self._build_enforced(adj_list)
if not self._manual_topo:
self._build_successors(adj_list, transition_adj_list)
self._build_long_dist_links(adj_list, transition_adj_list)
self._build_ondemand_links(adj_list, transition_adj_list, request_list)
for _, ce in adj_list.conn_edges.items():
assert ce.edge_state == "CEStateInitialized", "Invalid CE edge state, CE={}".format(ce)
return adj_list
def build_adj_list_ata(self,):
"""
Generates a new adjacency list from the list of available peers
"""
adj_list = ConnEdgeAdjacenctList(self.overlay_id, self._node_id,
self._max_successors, self._max_ldl_cnt, self._max_ond)
for peer_id in self._peers:
if self._enforced_edges and peer_id in self._enforced_edges:
ce = ConnectionEdge(peer_id)
ce.edge_type = "CETypeEnforced"
adj_list.add_conn_edge(ce)
elif not self._manual_topo and self._node_id < peer_id:
ce = ConnectionEdge(peer_id)
ce.edge_type = "CETypeSuccessor"
adj_list.add_conn_edge(ce)
return adj_list
def _distance(self, peer_id):
dst = 0
nsz = max(1, len(self._nodes))
try:
pr_i = self._nodes.index(peer_id)
dst = (pr_i + nsz - self._my_idx) % nsz
except ValueError as er:
self._top.log("LOG_WARNING", "%s, continuing ...", str(er))
return dst
def _ideal_closest_distance(self):
nsz = max(1, len(self._nodes))
off = math.exp(-1 * math.log10(nsz))
return math.floor(nsz * off)
def is_too_close(self, peer_id):
return self._distance(peer_id) < self._ideal_closest_distance()
def _prep(self, peers):
self._peers = peers
self._nodes = list(self._peers)
self._nodes.append(self._node_id)
self._nodes.sort()
self._my_idx = self._nodes.index(self._node_id)