I have two computers: Ubuntu1 and Ubuntu2. Ubuntu1 runs MongoDB with database Sacred3. I want to connect from U2 to U1 via ssh and store there my experiment results.
What I tried and failed:
1. I installed mongo DB, created sacred3, I have ssh key to it.
I edited /etc/mongod.conf
adding:
# network interfaces
net:
port: 27017
bindIp: 0.0.0.0
Then I enabled port forwarding with
ssh -fN -i ~/.ssh/sacred_key-pair.pem -L 6666:localhost:27017 ubuntu@106.969.696.969
// (with proper ip)
so, as I undertstand, if I connect to my localhost:6666 it will be forwarded to 106.969.696.969:27017
So after that, I'm runnig an experiment with Sacred framework:
python exp1.py -m localhost:6666:sacred3
and this should write experiment to remote DB, HOWEVER i I get:
pymongo.errors.ServerSelectionTimeoutError: localhost:27017: [Errno 111] Connection refused
which is driving me mad. please help!
#below contents of exp1.py:
from sacred import Experiment
from sacred.observers import MongoObserver
ex = Experiment()
ex.observers.append(MongoObserver.create())
def compute():
summ = layer1 - layer2
return summ
@ex.config
def my_config():
hp_list = [{"neurons" : [32,32] , "dropout": 1.0},
{"neurons" : [32,32] , "dropout": 0.7},
{"neurons" : [32,16] , "dropout": 0.9},
{"neurons" : [24,16] , "dropout": 0.9},
{"neurons" : [24,8] , "dropout": 0.9},
{"neurons" : [16,8] , "dropout": 0.9},
{"neurons" : [64,64] , "dropout": 0.9},
{"neurons" : [64,64] , "dropout": 0.7},
{"neurons" : [64,32] , "dropout": 0.9},
{"neurons" : [64,32] , "dropout": 0.7},
{"neurons" : [48,32] , "dropout": 0.9},
{"neurons" : [48,32] , "dropout": 0.7},
{"neurons" : [48,16] , "dropout": 0.9},
{"neurons" : [48,16] , "dropout": 0.7},]
n_epochs = 2
@ex.capture
def training_loop(hp_list, n_epochs):
for j in hp_list:
print("Epoch: ", n_epochs)
# layer1 = random.randint(18,68)
# layer2 = random.randint(18,68)
# layer3 = random.randint(18,68)
layer1 = j["neurons"][0]
layer2 = j["neurons"][1]
dropout_ratio = j["dropout"]
print("WHATS UUUUUP",j, layer1, layer2, dropout_ratio, sep="_")
# vae_training_loop_NN_DO(i, layer1, layer2, dropout_ratio )
@ex.automain
def my_main():
training_loop()