1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
|
# frozen_string_literal: true
module HTTPX
module Plugins
# This plugin implements a session that persists connections over the duration of the process.
#
# This will improve connection reuse in a long-running process.
#
# One important caveat to note is, although this session might not close connections,
# other sessions from the same process that don't have this plugin turned on might.
#
# This session will still be able to work with it, as if, when expecting a connection
# terminated by a different session, it will just retry on a new one and keep it open.
#
# This plugin is also not recommendable when connecting to >9000 (like, a lot) different origins.
# So when you use this, make sure that you don't fall into this trap.
#
# https://gitlab.com/os85/httpx/wikis/Persistent
#
module Persistent
class << self
def load_dependencies(klass)
klass.plugin(:fiber_concurrency)
max_retries = if klass.default_options.respond_to?(:max_retries)
[klass.default_options.max_retries, 1].max
else
1
end
klass.plugin(:retries, max_retries: max_retries)
end
end
def self.extra_options(options)
options.merge(persistent: true)
end
module InstanceMethods
def close(*)
super
# traverse other threads and unlink respective selector
# WARNING: this is not thread safe, make sure that the session isn't being
# used anymore, or all non-main threads are stopped.
Thread.list.each do |th|
store = thread_selector_store(th)
next unless store && store.key?(self)
selector = store.delete(self)
selector_close(selector)
end
end
private
def retryable_request?(request, response, *)
super || begin
return false unless response && response.is_a?(ErrorResponse)
error = response.error
Retries::RECONNECTABLE_ERRORS.any? { |klass| error.is_a?(klass) }
end
end
def retryable_error?(ex, options)
super &&
# under the persistent plugin rules, requests are only retried for connection related errors,
# which do not include request timeout related errors. This only gets overriden if the end user
# manually changed +:max_retries+ to something else, which means it is aware of the
# consequences.
(!ex.is_a?(RequestTimeoutError) || options.max_retries != 1)
end
end
end
register_plugin :persistent, Persistent
end
end
|
