I think this is more of a Kazoo lock recipe question and not ZooKeeper
specific, but I did some research for you:
It seems that Kazoo locks are based on the path parameter that you specify
when creating a new lock object and _not_ the identifier (name in your
case). So basically second call (should) timeouts, because you're trying to
acquire the same lock 10 times. Changing the code to use PATH as a prefix
and concatenate with the random name fixes it:
PATH = "aggregator_test/"
lock = zoo_conn.Lock(PATH + name)
On Wed, Jan 31, 2018 at 1:10 PM, Alexei <dm...@yandex.ru> wrote:
> We have 3 nodes Zoo cluster running 3.4.5+dfsg-1 on ubuntu 14. Client is
> python with 1.2.1 and (tried) 2.4.0 python-kazoo
> When I run simple script which set lock in loop it starts to time out
> after 1 or 2 set locks.
> What am I doing wrong?
> Script is here: http://paste.org.ru/?xqlivn
> Log is here: http://paste.org.ru/?zwr6u1
> root@zk-gce-eu-2:~# echo mntr | nc -v localhost 2181
> Connection to localhost 2181 port [tcp/*] succeeded!
> zk_version 3.4.5--1, built on 06/10/2013 17:26 GMT
> zk_avg_latency 0
> zk_max_latency 203
> zk_min_latency 0
> zk_packets_received 2741185
> zk_packets_sent 2963217
> zk_num_alive_connections 24
> zk_outstanding_requests 0
> zk_server_state leader
> zk_znode_count 51
> zk_watch_count 5
> zk_ephemerals_count 44
> zk_approximate_data_size 4770
> zk_open_file_descriptor_count 68
> zk_max_file_descriptor_count 8192
> zk_followers 2
> zk_synced_followers 2
> zk_pending_syncs 0
> Thank you.
> Grats, Alex.