Check failed: mdb_status == 0 (12 vs. 0) Cannot allocate memory

Question

I am creating an lmdb database with following settings:

train_lmdb = lmdb.open(train_lmdb_name, map_size=int(1e13), writemap=True)
lmdb_txn = train_lmdb(write=True)

My input data is of shape 128x128x128 and I have 900.000 samples. Creating the database is no problem, but when trying to train the dataset using caffe I receive following error:

Check failed: mdb_status == 0 (12 vs. 0) Cannot allocate memory

I think it is because I use map_size=int(1e13) rather than 1e12 and therefore I get the error, however my dataset is 128x128x128x900000 = 1.887.436.800.000 > 1e12 what can I do?

Stacktrace:

I0827 12:00:06.379778 11729 net.cpp:198] conv1_b_conv1_b_relu_0_split needs backward computation.
I0827 12:00:06.379784 11729 net.cpp:198] conv1_b_relu needs backward computation.
I0827 12:00:06.379789 11729 net.cpp:198] scale_conv1_b needs backward computation.
I0827 12:00:06.379794 11729 net.cpp:198] bn_conv1_b needs backward computation.
I0827 12:00:06.379799 11729 net.cpp:198] conv1_b needs backward computation.
I0827 12:00:06.379806 11729 net.cpp:200] volume does not need backward computation.
I0827 12:00:06.379811 11729 net.cpp:200] data does not need backward computation.
I0827 12:00:06.379814 11729 net.cpp:242] This network produces output linear2
I0827 12:00:06.379820 11729 net.cpp:242] This network produces output loss
I0827 12:00:06.380133 11729 net.cpp:255] Network initialization done.
I0827 12:00:06.380894 11729 solver.cpp:56] Solver scaffolding done.
I0827 12:00:06.400212 11729 caffe.cpp:248] Starting Optimization
F0827 12:00:07.560200 11766 db_lmdb.hpp:15] Check failed: mdb_status == 0 (12 vs. 0) Cannot allocate memory
*** Check failure stack trace: ***
    @     0x2b0d2be31b2d  google::LogMessage::Fail()
    @     0x2b0d2be33995  google::LogMessage::SendToLog()
    @     0x2b0d2be316a9  google::LogMessage::Flush()
    @     0x2b0d2be3442e  google::LogMessageFatal::~LogMessageFatal()
    @     0x2b0d2ac7aede  caffe::db::LMDB::Open()
    @     0x2b0d2ab67667  caffe::DataLayer<>::DataLayer()
    @     0x2b0d2ab67922  caffe::Creator_DataLayer<>()
    @     0x2b0d2abfca1b  caffe::LayerRegistry<>::CreateLayer()
    @     0x2b0d2ac3c97a  caffe::Net<>::Init()
    @     0x2b0d2ac3ed55  caffe::Net<>::Net()
    @     0x2b0d2ac4fdf6  caffe::Solver<>::InitTrainNet()
    @     0x2b0d2ac51363  caffe::Solver<>::Init()
    @     0x2b0d2ac5167f  caffe::Solver<>::Solver()
    @     0x2b0d2ac62301  caffe::Creator_AdamSolver<>()
    @           0x415c6c  caffe::SolverRegistry<>::CreateSolver()
    @     0x2b0d2ac4b32f  caffe::Worker<>::InternalThreadEntry()
    @     0x2b0d2ab01185  caffe::InternalThread::entry()
    @     0x2b0d2ab01b6e  boost::detail::thread_data<>::run()
    @     0x2b0d2b3e2739  thread_proxy
    @     0x2b0d2c6cedc5  start_thread
    @     0x2b0d414c873d  __clone