def report_monitor(node_ip, osd_id):
flag = 0
#Will call monitor to state about the down node
soc = socket.socket()
soc.settimeout(5)
print ("Socket successfully created for Recovery: Primary")
monitor_1 = monitor_ip["primary"]
try :
soc.connect((monitor_1["ip"], monitor_1["port"]))
# soc.timeout(None)
print(f"Connecting Primary monitor...")
res = {"type": "FAIL", "ip" : node_ip, "osd_id" : osd_id}
_send_msg(soc, res)
msg = _wait_recv_msg(soc, MSG_SIZE)
print(msg)
if msg == None:
pass
elif msg["type"] == "ACK":
soc.close()
return
except Exception as e:
print(e)
print("Didn't Connect! [Timeout] Primary Monitor is Down")
soc.close()
soc = socket.socket()
soc.settimeout(5)
print ("Socket successfully created for Recovery: Backup")
monitor_2 = monitor_ip["backup"]
try :
soc.connect((monitor_2["ip"], monitor_2["port"]))
soc.settimeout(None)
print(f"Connecting Backup monitor...")
res = {"type": "FAIL", "ip" : node_ip, "osd_id" : osd_id}
_send_msg(soc, res)
msg = _wait_recv_msg(soc, MSG_SIZE)
print(msg)
if msg == None:
pass
elif msg["type"] == "ACK":
return
except:
print("MAY GOD HELP US!! WE ARE DOOMED\n\n")
soc.close()
def logout(self):
msg = {"type": "CLIENT_LOGOUT", "username": self.username}
s = socket.socket()
# s.bind(('', 9090))
ip = MDS_IPs["primary"]["ip"]
port = MDS_IPs["primary"]["port"]
try:
s.connect((ip, port))
_send_msg(s, msg)
response = _wait_recv_msg(s, MSG_SIZE)
if response["status"] == "SUCCESS": # here check response from mds
self.window.destroy()
print(response["msg"])
elif response["status"] == "ERROR":
print(response["msg"])
except Exception as e:
print("[ERROR] login failed")
print(e)
finally:
global LoginPage
s.close()
LoginPage()
def _read(self, pg_id):
s = socket.socket()
# send monitor to ask for OSD details
ip = MONITOR_IPs["primary"]
port = CLIENT_REQ_PORT
s.connect((ip, port))
msg = {"type": "READ", "pg_id": pg_id}
# d_msg = pickle.dumps(msg)
_send_msg(s, msg)
#s.send(d_msg)
response = _wait_recv_msg(s, 1024)
osd_addr = response["addrs"][0]
# print(osd_ips)
s.close()
# print(osd_ips[0][0], osd_ips[0][1])
# write on OSD
s = socket.socket()
s.connect(osd_addr)
data_msg = {"type": "READ", "pg_id": pg_id}
_send_msg(s, data_msg)
osd_response = _wait_recv_msg(s, 1024)
# print(osd_response)
# print(osd_response["pg_id"] == pg_id)
# print(osd_response["res"] == "SUCCESS")
if osd_response["pg_id"] == pg_id and osd_response["res"] == "SUCCESS":
s.close()
print("PG received from OSD writing on disk")
return 0, osd_response["pg"]
else:
s.close()
print("Error - PG not received from OSD")
return -2, None
def recv_gossip():
while True:
time.sleep(10)
# Wait for 10 sec to run this protocol
i = 0
for i in range(4):
curr_osd = "osd_id" + str(i + 1)
if curr_osd != STORAGE_ID:
node_ip = storage_ip[curr_osd]["ip"]
port = storage_ip[curr_osd]["port"]
soc = socket.socket()
soc.settimeout(5)
print(
f"\n\nSocket successfully created for Gossip with osd {curr_osd}"
)
try:
soc.connect((node_ip, port))
soc.settimeout(None)
print(
f"Connecting {node_ip} storage node number {curr_osd} port {port}"
)
msg = {"type": "ALIVE"}
_send_msg(soc, msg)
rec = _wait_recv_msg(soc, MSG_SIZE)
print(msg)
if rec == None:
print(
f"Didn't receive data to Storage {curr_osd} ip {node_ip}! [Timeout] "
)
report_monitor(node_ip, curr_osd)
elif rec["type"] != "ACK":
report_monitor(node_ip, curr_osd)
except:
print(
f"Didn't Connect to Storage {curr_osd} ip {node_ip}! [Timeout]"
)
report_monitor(node_ip, curr_osd)
soc.close()
time.sleep(3)
def login():
global user, passwd, login_screen
# send credentials to mds and wait for response
client_id = "C200" # received from mds
print("username:: ", user.get())
# print("password:: ", passwd.get())
msg = {
"type": "CLIENT_LOGIN",
"username": user.get(),
"password": passwd.get()
}
s = socket.socket()
ip = MDS_IPs["primary"]["ip"]
port = MDS_IPs["primary"]["port"]
# print(s.gethostname())
# s.bind(('', 9090))
try:
s.connect((ip, port))
_send_msg(s, msg)
response = _wait_recv_msg(s, MSG_SIZE)
s.close()
if response["status"] == "SUCCESS": # here check response from mds
print(response["msg"])
login_screen.destroy()
client = Client(response["tree"])
elif response["status"] == "ERROR":
# print(response["msg"])
_popup("Login Failed", response["msg"])
except Exception as e:
s.close()
print("[ERROR] login failed")
print(e)
_popup("Login Failed", str(e))
finally:
print("Exiting login..")
def update_backup_monitor(update_type, pg_or_osd_ids_list, osd_list):
primary_update_socket = socket.socket()
print ("primary update socket successfully created")
try:
primary_update_socket.connect((MONITOR_IPs["backup"], RECV_PRIMARY_UPDATE_PORT))
msg = {"update_type": update_type, "pg_or_osd_ids_list": pg_or_osd_ids_list, \
"osd_list": osd_list}
_send_msg(primary_update_socket, msg)
response = _wait_recv_msg(primary_update_socket, MSG_SIZE)
primary_update_socket.close()
if response["status"] == "SUCCESS":
print("written to monitor backup successfully")
else:
print("write to monitor backup failed")
except Exception as e:
print(e)
primary_update_socket.close()
def _update_primary(self, update_type, update):
print("writing to backup..")
s = socket.socket()
# send backup mds to make data consistent
ip = MDS_IPs["backup"]["ip"]
port = MDS_IPs["backup"]["port"]
try:
s.connect((ip, port))
msg = {"type": "UPD", "update_type": update_type, "update": update}
# d_msg = pickle.dumps(msg)
_send_msg(s, msg)
#s.send(d_msg)
response = _wait_recv_msg(s, 1024)
if response["status"] == "SUCCESS":
if update_type == "ADD_LOGIN_USER":
self.logged_in.append(update["user"])
self._write_logged_in()
elif update_type == "REMOVE_LOGIN_USER":
self.logged_in.remove(update["user"])
self._write_logged_in()
print("Successfully updated on Backup")
return 0
elif response["status"] == "ERROR":
print(response["msg"])
print("Couldn't update on Backup")
return -1
except Exception as e:
print(e)
print("Couldn't update on Backup")
return -2
finally:
s.close()
print("writing to backup completed.")
def send_replicate_request(pg_id, master_osd, clone_osd):
soc = socket.socket()
soc.settimeout(5)
print(
f"Socket successfully created for replicating {pg_id} from master {master_osd} to clone {clone_osd}"
)
# reserve a port on your computer in our
ip_add = storage_ip[master_osd]["ip"]
port = storage_ip[master_osd]["port"] + 10
flag = -1
try:
soc.connect((ip_add, port))
soc.settimeout(None)
print(f"Connecting with {master_osd} on port {port}...")
res = {"type": "REPLICATE", "pg_id": pg_id, "osd_id": clone_osd}
_send_msg(soc, res)
print("Message send for replication ")
# time.sleep(5)
msg = _wait_recv_msg(soc, MSG_SIZE)
print(msg)
if msg == None:
flag = 0
elif msg["type"] == "ACK":
flag = 1
except Exception as e:
print(e)
print(
f"Didn't Connect! [Timeout] master {master_osd} is down, port {port}"
)
soc.close()
return flag
ip = MDS_IPs["primary"]["ip"]
port = MDS_IPs["primary"]["port"]
print("receiving data..")
s = socket.socket()
try:
s.connect((ip, port))
msg = {"type":"RECOVERY"}
# d_msg = pickle.dumps(msg)
_send_msg(s, msg)
#s.send(d_msg)
response = _wait_recv_msg(s, 1024)
if response["status"] == "SUCCESS":
trees = response["trees"]
logged_in = response["logged_in"]
# user_list = response["user_list"]
for tree in trees:
file = open("./tree/"+tree["username"], 'wb')
obj_b = pickle.dumps(tree)
file.write(obj_b)
file.close()
file = open("./logged_in", 'wb')
def replicate_pg(soc):
# This will be always true loop that will run in parallel
c, addr = soc.accept()
print(f"Got connection from {addr} to do the replication")
n = os.fork()
if n == 0:
msg = _recv_msg(c, MSG_SIZE)
print(msg)
if msg == None:
print(f"Didn't receive data! [Timeout] {addr}")
msg = {"type": "SEND AGAIN"}
_send_msg(c, msg)
elif msg["type"] == "REPLICATE":
# Can do one thing here
# call the func that act as access this pg_id for the client
# then call other function that act as send this pg to osd to do a write
pg_id = msg["pg_id"]
osd_id = msg["osd_id"]
file = open("./data/" + pg_id, 'rb')
pg_b = file.read()
pg = pickle.loads(pg_b)
## Connect to this osd_id with new socket
new_soc = socket.socket()
print(
f"Socket created to send request to SAVE data at OSD {osd_id}")
ip_add = storage_ip[osd_id]["ip"]
port = storage_ip[osd_id]["port"] + 10
# Remember replication is on normal port + 10
try:
new_soc.connect((ip_add, port))
print(f"Connection made with {ip_add} on {port}")
msg = {"type": "SAVE", "pg_id": pg_id, "pg": pg}
_send_msg(new_soc, msg)
print("Msg send to SAVE the data")
msg = _wait_recv_msg(new_soc, MSG_SIZE)
print("Msg received !")
except Exception as e:
print(e)
if msg == None:
res = {"type": "REPLICATION FAIL"}
_send_msg(c, res)
print("Fail Msg send to the monitor")
elif msg["type"] == "ACK":
res = {"type": "ACK"}
_send_msg(c, res)
print("Success Msg send to the monitor")
elif msg["type"] == "SAVE":
# This will take request from other osd to save some data
# This is basic replication strategy
pg_id = msg["pg_id"]
pg = msg["pg"]
try:
file = open("./data/" + pg_id, 'wb')
print("Replicated data is saved")
pg_dump = pickle.dump(pg)
file.write(pg_dump)
except Exception as e:
print(e)
msg = {"type": "ACK"}
_send_msg(c, msg)
print("Write successful..send back Ack to the master osd")
else:
print("[ERROR] Check the code in replication")
c.close()
print(f"Exiting from pid {os.getpid()} ..")
os._exit(1)
def update_handler(self):
# if n == 0:
print("update handler started..")
# if self.start_update_handler == True:
while True:
if len(self.processing) == 0:
self.start_update_handler = False
return
msg = {
"type": "WRITE_QUERY",
"username": self.username,
"processing": self.processing
}
s = socket.socket()
ip = MDS_IPs["primary"]["ip"]
port = MDS_IPs["primary"]["port"]
# print(s.gethostname())
# s.bind(('', 9090))
try:
s.connect((ip, port))
_send_msg(s, msg)
response = _wait_recv_msg(s, MSG_SIZE)
s.close()
if response[
"status"] == "SUCCESS": # here check response from mds
print(response["msg"])
file_written = response["file_written"]
for file in file_written: #populate listbox again
filename = file[0]
file_id = file[1]
file_path = file[2]
self.file_id_map[filename] = file_id
self.listbox.insert(tk.END, file_path)
# listbox.insert(END, percent)
self.listbox.update_idletasks()
tree = response["tree"]
self.dir_tree = tree["dir_tree"]
self.processing = tree["processing"]
print(self.processing)
if len(self.processing) == 0:
self.start_update_handler = False
print("Exiting update handler")
for file in file_written:
self._popup("Update",
str(file[2]) + " UPLOADED SUCCESSFULLY")
# return
# update on GUI
# self._update_gui()
# for filename in file_written: #populate listbox again
# self.listbox.insert(END, filename)
# # listbox.insert(END, percent)
# self.listbox.update_idletasks()
# tk.mainloop()
# self.window.mainloop()
elif response["status"] == "NO_UPD":
print(response["msg"])
time.sleep(self.update_interval)
except Exception as e:
s.close()
print("Update (write) from MDS failed")
print(e)
time.sleep(self.update_interval)
finally:
time.sleep(self.update_interval)
def _write(self, pg, pg_data):
s = socket.socket()
# send monitor to ask for OSD details
ip = MONITOR_IPs["primary"]
port = CLIENT_REQ_PORT
s.connect((ip, port))
msg = {
"type": "WRITE",
"pg_id": pg.pg_id,
"size": (sys.getsizeof(pg) / float(1 << 20))
}
# d_msg = pickle.dumps(msg)
_send_msg(s, msg)
#s.send(d_msg)
res = _wait_recv_msg(s, MSG_SIZE)
osd_dict = {}
if res["status"] == "SUCCESS":
osd_dict = res["osd_dict"]
else:
print(res["msg"])
s.close()
return -2
# osd_dict = response["osd_dict"]
# print(osd_ips)
s.close()
# print(osd_ips[0][0], osd_ips[0][1])
# write on OSD
osd_addr = osd_dict["addrs"][osd_dict["osd_ids"][0]]
s = socket.socket()
s.connect(osd_addr)
data_msg = {
"type": "CLIENT_WRITE",
"pg": pg,
"size": (sys.getsizeof(pg) / float(1 << 20)),
"client_id": self.username,
"client_addr": "",
"osd_dict": osd_dict
}
_send_msg(s, data_msg)
print(len(pg.object_list))
self.processing[pg.pg_id] = pg_data
threading.Thread(target=self.update_handler()).start()
### Add server to receive response
osd_response = _wait_recv_msg(s, 1024)
if osd_response["status"] == "RECEIVED":
print("file sent to OSD")
# self.processing[pg.pg_id] = pg_data
# if self.start_update_handler == False:
# self.start_update_handler = True
# self.update_handler()
# threading.Thread(target=self.update_handler())
s.close()
return 0
else:
s.close()
print("Error - File not sent")
return -2
def recv_write_acks():
global hashtable, cluster_topology, MDS_IP, MDS_flags
write_ack_socket = socket.socket()
print ("write ack socket successfully created")
write_ack_socket.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
# reserve a port on your computer
port = WRITE_ACK_PORT
# Next bind to the port
# we have not entered any ip in the ip field
# instead we have inputted an empty string
# this makes the server listen to requests
# coming from other computers on the network
write_ack_socket.bind(('', port))
print ("write ack socket bound to %s" %(port))
# put the socket into listening mode
write_ack_socket.listen(5)
print ("socket is listening")
# a forever loop until we interrupt it or
# an error occurs
while True:
# Establish connection with osd
c, addr = write_ack_socket.accept()
print ('Got connection from', addr)
# recv the acknowledgement
ack = _recv_msg(c, 1024)
print(ack)
# extracting the written pd_id, free_space, osd_id of that osd
client_id = ack["client_id"]
# client_addr = ack["client_addr"] # addr = (ip, port)
pg_id = ack["pg_id"]
free_space = ack["free_space"]
osd_id = ack["osd_id"]
# if the osd is down or out, make it up again
# [need to check this]
if cluster_topology[osd_id]["status"] != 0:
cluster_topology[osd_id]["status"] = 0
# updating the write status of the osd to 1
for i in range(len(hashtable[pg_id])):
if hashtable[pg_id][i][0] == osd_id:
hashtable[pg_id][i][1] = 1
replication_factor = 0
# adding the new friend
for osd in hashtable[pg_id]:
if osd[0] != osd_id:
if osd[1] == 1:
replication_factor += 1
cluster_topology[osd_id]["friends"].add(osd[0])
cluster_topology[osd[0]]["friends"].add(osd_id)
cluster_topology[osd_id]["free_space"] = free_space
if isPrimary:
# updating the backup
update_backup_monitor("hash_table", [pg_id], [hashtable[pg_id]])
update_backup_monitor("cluster_topology", [osd[0] for osd in hashtable[pg_id]], \
[cluster_topology[osd[0]] for osd in hashtable[pg_id]])
hashtable_file = open('hashtable', 'wb')
cluster_topology_file = open('cluster_topology', 'wb')
hashtable_dump = pickle.dumps(hashtable)
hashtable_file.write(hashtable_dump)
hashtable_file.close()
cluster_topology_dump = pickle.dumps(cluster_topology)
cluster_topology_file.write(cluster_topology_dump)
cluster_topology_file.close()
if replication_factor > 1:
# # sending write update to client
# client_update = socket.socket()
# print ("client write ack socket successfully created")
# client_update.connect(client_addr)
# msg = {"type": "WRITE_RESPONSE", "PG_ID": pg_id, \
# "status": "SUCCESS", "message": "write successful",\
# "client_id": client_id}
# _send_msg(client_update, msg)
# client_update.close()
# sending write update to MDS
MDS_update_socket = socket.socket()
print ("MDS write ack socket successfully created")
MDS_update_socket.connect((MDS_IP, MDS_PORT))
msg = {"type": "WRITE_RESPONSE", "PG_ID": pg_id, \
"status": "SUCCESS", "message": "write successful",\
"client_id": client_id}
_send_msg(MDS_update_socket, msg)
MDS_ack = _wait_recv_msg(MDS_update_socket, 1024)
pg_id = MDS_ack["pg_id"]
if MDS_ack["status"] == "SUCCESS":
print("Write successful")
MDS_flags[pg_id] = 0
else:
print("Write error from MDS")
print(MDS_ack["msg"])
MDS_flags[pg_id] = 1
## resend pgs in MDS_flags with flag = 1
## start that loop here
MDS_update_socket.close()
error = ""
status = "SUCCESS"
# send success
_send_msg(c, {"error":error, "status":status})
c.close()
write_ack_socket.close()