elasticsearch-ruby icon indicating copy to clipboard operation
elasticsearch-ruby copied to clipboard

Stuck on loop because the number of connections keeps increasing

Open picandocodigo opened this issue 3 years ago • 1 comments

Has anybody had this issue where it continues looping forever because the number of connections keeps increasing?

Error -- 3 -- [Faraday::TimeoutError] request timed out {:id=>"kbnoHraTQN6xu2cgbUTeOQ", :name=>"production-master2", :version=>"5.6.16", :host=>"10.0.11.4", :port=>"9200", :roles=>["master"], :attributes=>nil, :protocol=>"http"}
Warning -- 4 -- [Faraday::TimeoutError] Reloading connections (attempt 1 of 14)
Error -- 3 -- [Faraday::TimeoutError] request timed out {:id=>"VYG1SZilTGuIS_FYOSKTRg", :name=>"production-master1", :version=>"5.6.16", :host=>"10.0.10.52", :port=>"9200", :roles=>["master"], :attributes=>nil, :protocol=>"http"}
Warning -- 4 -- [Faraday::TimeoutError] Reloading connections (attempt 2 of 15)
Fatal -- 1 -- [IRB::Abort] abort then interrupt! ({:id=>"Aei_vKRMQJSm2pK3Xphp2Q", :name=>"production-data-ebs1", :version=>"5.6.16", :host=>"10.0.10.49", :port=>"9200", :roles=>["data", "ingest"], :attributes=>nil, :protocol=>"http"})

1/14 2/15 3/16 etc

It seems dead connections get somehow duplicated of existing alive ones. I can see essentially the same connection (same host, post, attributes, etc) both in the dead set and alive one (different instances of the connection, not the same).

I've added debug logs to __rebuild_connections and this is what I got:

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 14
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 15

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 15
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 16

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 16
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 17

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 17
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 18

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 18
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 19

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 19
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 20

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 20
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 21

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 21
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 22

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 22
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 23

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 23
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 24

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 24
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 25

Hosts count: 14
new_connections size: 14
stale_connections size: 0
new_connections (after reject) size: 1
@connections.all size: 25
@connections size: 13
@connections (after) size: 14
@connections.all (after) size: 26

If I'm correct, we need to replace

new_connections = new_connections.reject { |c| @connections.include?(c) }

with

new_connections = new_connections.reject { |c| @connections.all.include?(c) }

Originally posted by @andrepcg in https://github.com/elastic/elasticsearch-ruby/issues/695#issuecomment-1056898221

picandocodigo avatar Apr 19 '22 07:04 picandocodigo