honey_pool is yet another gun connection pool based on worker pool.
Alpha Quality: Things can change.
Consider using honey_pool when your hackney_pool hangs while sending massive amount of HTTP reqs. honey_pool starts multiple connection pools using worker pool, and manages gun HTTP connections in the pools.
In the rebar.config deps, have:
{honey_pool, {git, "git://github.com/yowcow/honey_pool.git", {branch, "master"}}}
and start application honey_pool in your app, then try in shell:
1> logger:set_primary_config(level, info).
ok
2> honey_pool:dump_state().
[#{await_up_conns => #{},host_conns => #{},up_conns => #{}},
#{await_up_conns => #{},host_conns => #{},up_conns => #{}},
#{await_up_conns => #{},host_conns => #{},up_conns => #{}}]
3> honey_pool:get("http://example.com/", 100).
{error,{checkout,{timeout,await_up}}}
4> honey_pool:dump_state().
[#{await_up_conns => #{},
host_conns =>
#{{"example.com",80,tcp} =>
[{<0.195.0>,#Ref<0.485877779.3106144257.39748>,
#Ref<0.485877779.3106144257.39749>}]},
up_conns => #{<0.195.0> => {"example.com",80,tcp}}},
#{await_up_conns => #{},host_conns => #{},up_conns => #{}},
#{await_up_conns => #{},host_conns => #{},up_conns => #{}}]
5> honey_pool:summarize_state().
[#{host_conns => [{{"example.com",80,tcp},1}],
total_conns => #{await_up => 0,up => 1}},
#{host_conns => [],total_conns => #{await_up => 0,up => 0}},
#{host_conns => [],total_conns => #{await_up => 0,up => 0}}]
In the sys.config, have:
{honey_pool, [
%% honey pool configurations
{idle_timeout, 60000}, %% close connection after 60 sec of idle
%% worker pool configurations (see worker pool docs for details)
{wpool, [
{workers, 10}, %% start 10 connection pools
{overrun_warning, 50} %% warn when a pool takes over 50 msec to respond
]},
%% gun configurations (see gun docs for details)
{gun_opt, #{
retry => 0, %% let go dead connections
connect_timeout => 1000 %% give up connecting after 1000 msec
}}
]}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent