So, while perusing lmdb.h I saw this:

Use an MDB_env* in the process which opened it, not after fork().

That was my problem. It had nothing to do with transactions. I need to open the environment and hold a separate handle in each process. Right?

Moving the mdb_env_create and mdb_env_open functions and related variables inside the get_() function got rid of the errors. In my dummy code below that is obviously terribly expensive, but my real application is running a pool of long-lived workers (HTTP server) and the cost of opening the environments only happens at server start.

Thanks, Stefano

On 08/19/2018 10:36 AM, Stefano Cossu wrote:

Hello, I am writing a framework in Cython using multi-process access to an LMDB database. I ran into a "Resource not available" error.

I isolated the problem in the following script (this is Cython calling the LMDB C API, so bear with the hybrid syntax):

import time cimport cylmdb as lmdb # This is a Cython header mirroring lmdb.h

import multiprocessing import threading

cdef:     lmdb.MDB_env *env     lmdb.MDB_dbi dbi

cdef void _check(int rc) except *:     if rc != lmdb.MDB_SUCCESS:         out_msg = 'LMDB Error ({}): {}'.format(             rc, lmdb.mdb_strerror(rc).decode())         raise RuntimeError(out_msg)

cpdef void get_() except *:     cdef:         lmdb.MDB_txn *txn         lmdb.MDB_val key_v, data_v

_check(lmdb.mdb_txn_begin(env, NULL, lmdb.MDB_RDONLY, &txn))

key_v.mv_data = b'a'     key_v.mv_size = 1

_check(lmdb.mdb_get(txn, dbi, &key_v, &data_v))     print((<unsigned char *>data_v.mv_data)[:data_v.mv_size])     time.sleep(1)     _check(lmdb.mdb_txn_commit(txn))     print('Thread {} in process {} done.'.format(             threading.currentThread().getName(),             multiprocessing.current_process().name))

def run():     cdef:         unsigned int flags = 0         #unsigned int flags = lmdb.MDB_NOTLS # I tried this too.         lmdb.MDB_txn *wtxn         lmdb.MDB_val key_v, data_v

# Set up environment.     _check(lmdb.mdb_env_create(&env))     _check(lmdb.mdb_env_set_maxreaders(env, 128))     _check(lmdb.mdb_env_open(env, '/tmp/test_mp', flags, 0o644))

# Create DB.     _check(lmdb.mdb_txn_begin(env, NULL, 0, &wtxn))     _check(lmdb.mdb_dbi_open(wtxn, NULL, lmdb.MDB_CREATE, &dbi))

# Write something.     key_v.mv_data = b'a'     key_v.mv_size = 1     ts = str(time.time()).encode()     data_v.mv_data = <unsigned char *>ts     data_v.mv_size = len(ts)     _check(lmdb.mdb_put(wtxn, dbi, &key_v, &data_v, 0))     _check(lmdb.mdb_txn_commit(wtxn))

print('Multiprocess jobs:')     for i in range(5):         multiprocessing.Process(target=get_).start()     # Env should be closed only after all processes return.     #lmdb.mdb_env_close(env)

If I execute run(), the first process runs successfully, but apparently it's holding on to some resources that the other processes need.

Multiprocess jobs: b'1534691578.4300401' Process Process-2: Traceback (most recent call last):   File "/usr/lib/python3.6/multiprocessing/process.py", line 258, in _bootstrap     self.run()   File "/usr/lib/python3.6/multiprocessing/process.py", line 93, in run     self._target(*self._args, **self._kwargs)   File "lakesuperior/sandbox/threading_poc.pyx", line 23, in lakesuperior.sandbox.threading_poc.get_     cpdef void get_() except *:   File "lakesuperior/sandbox/threading_poc.pyx", line 28, in lakesuperior.sandbox.threading_poc.get_     _check(lmdb.mdb_txn_begin(env, NULL, lmdb.MDB_RDONLY, &txn))   File "lakesuperior/sandbox/threading_poc.pyx", line 20, in lakesuperior.sandbox.threading_poc._check     raise RuntimeError(out_msg) RuntimeError: LMDB Error (11): Resource temporarily unavailable Process Process-3: Traceback (most recent call last):   File "/usr/lib/python3.6/multiprocessing/process.py", line 258, in _bootstrap     self.run()   File "/usr/lib/python3.6/multiprocessing/process.py", line 93, in run     self._target(*self._args, **self._kwargs)   File "lakesuperior/sandbox/threading_poc.pyx", line 23, in lakesuperior.sandbox.threading_poc.get_     cpdef void get_() except *:   File "lakesuperior/sandbox/threading_poc.pyx", line 28, in lakesuperior.sandbox.threading_poc.get_     _check(lmdb.mdb_txn_begin(env, NULL, lmdb.MDB_RDONLY, &txn))   File "lakesuperior/sandbox/threading_poc.pyx", line 20, in lakesuperior.sandbox.threading_poc._check     raise RuntimeError(out_msg) RuntimeError: LMDB Error (11): Resource temporarily unavailable Process Process-4:

Process Process-5: Traceback (most recent call last):   File "/usr/lib/python3.6/multiprocessing/process.py", line 258, in _bootstrap     self.run()   File "/usr/lib/python3.6/multiprocessing/process.py", line 93, in run     self._target(*self._args, **self._kwargs)   File "lakesuperior/sandbox/threading_poc.pyx", line 23, in lakesuperior.sandbox.threading_poc.get_     cpdef void get_() except *: Traceback (most recent call last):   File "lakesuperior/sandbox/threading_poc.pyx", line 28, in lakesuperior.sandbox.threading_poc.get_     _check(lmdb.mdb_txn_begin(env, NULL, lmdb.MDB_RDONLY, &txn))   File "lakesuperior/sandbox/threading_poc.pyx", line 20, in lakesuperior.sandbox.threading_poc._check     raise RuntimeError(out_msg)   File "/usr/lib/python3.6/multiprocessing/process.py", line 258, in _bootstrap     self.run()   File "/usr/lib/python3.6/multiprocessing/process.py", line 93, in run     self._target(*self._args, **self._kwargs)   File "lakesuperior/sandbox/threading_poc.pyx", line 23, in lakesuperior.sandbox.threading_poc.get_     cpdef void get_() except *: RuntimeError: LMDB Error (11): Resource temporarily unavailable   File "lakesuperior/sandbox/threading_poc.pyx", line 28, in lakesuperior.sandbox.threading_poc.get_     _check(lmdb.mdb_txn_begin(env, NULL, lmdb.MDB_RDONLY, &txn))   File "lakesuperior/sandbox/threading_poc.pyx", line 20, in lakesuperior.sandbox.threading_poc._check     raise RuntimeError(out_msg) RuntimeError: LMDB Error (11): Resource temporarily unavailable

If I run the same function multiple times on the same process, everything is fine.

Can someone point out what is wrong with this script?

Thank you, Stefano