Kazoo hangs in multiprocessing

[ZedMcChen]

The following contrived multiprocessing code hangs when run. Kazoo seems to be waiting for its async_set to complete.

#!/usr/bin/python3 from kazoo.client import KazooClient from multiprocessing import Pool

def init_zk_conn(zk_server): kz = KazooClient(zk_server, timeout=10) kz.start() return kz

def update_znode(kz, znode, msg=''): if not kz.exists(znode): kz.create(znode, value=msg.encode('utf-8')) else: kz.set(znode, msg.encode('utf-8'))

KZ = init_zk_conn('localhost:2181')

def update_node(node): znode = '/testkz/' + node update_znode(KZ, znode, 'updating')

if name == 'main': pool = Pool(2) node_list = ['one', 'two'] pool.map(update_node, node_list) print('Done')

[ShireCiel]

I found init Kazoo before fork, and reuse it after fork it will hang. It's looks similar. You may can try (do not init any kazoo client before using and )init their own kazoo in their own process

[hungrybirder]

Any progress on this issue?