Client allows endless loops if there is a network issue

Algolia Client Version: Only tested 1.26.0
Language Version: Not relevant.

Description

This is basically an endless loop:

algoliasearch-client-ruby/lib/algolia/client.rb

Lines 494 to 511 in 55e2fb9

    
           # 
        
           # Wait the publication of a task on the server. 
        
           # All server task are asynchronous and you can check with this method that the task is published. 
        
           # 
        
           # @param index_name the index name owning the taskID 
        
           # @param taskID the id of the task returned by server 
        
           # @param time_before_retry the time in milliseconds before retry (default = 100ms) 
        
           # @param request_options contains extra parameters to send with your query 
        
           # 
        
           def wait_task(index_name, taskID, time_before_retry = WAIT_TASK_DEFAULT_TIME_BEFORE_RETRY, request_options = {}) 
        
             loop do 
        
               status = get_task_status(index_name, taskID, request_options) 
        
               if status == 'published' 
        
                 return 
        
               end 
        
               sleep(time_before_retry.to_f / 1000) 
        
             end 
        
           end

There should be some retry logic, and if enough fails it should throw an error.

Related: when perform a save_objects! with an empty batch, the taskID returned from the engine never resolves, resulting in this loop running forever.

client = ::Algolia::Search::Client.create_with_config(
::Algolia::Search::Config.new(
application_id: ENV['ALGOLIA_APP_ID'],
api_key: ENV['ALGOLIA_ADMIN_API_KEY'],
batch_size: ENV['ALGOLIA_BATCH_SIZE']
)
)

index = client.init_index('test-index-name')
index.save_objects!([]) # Hangs forever

We'll need to come up with a proper way to break here after x loops to prevent the infinite loop. I'll discuss this internally to see what we want to do about it.

thanks for looking into the issue : )

This code still seem to be causing endless loops when there's issues on Algolias servers.

https://status.algolia.com/incidents/4367

@chuckmeyer reached out to me on Twitter, and pointed me to this PR: #487

	#
	# Wait the publication of a task on the server.
	# All server task are asynchronous and you can check with this method that the task is published.
	#
	# @param index_name the index name owning the taskID
	# @param taskID the id of the task returned by server
	# @param time_before_retry the time in milliseconds before retry (default = 100ms)
	# @param request_options contains extra parameters to send with your query
	#
	def wait_task(index_name, taskID, time_before_retry = WAIT_TASK_DEFAULT_TIME_BEFORE_RETRY, request_options = {})
	loop do
	status = get_task_status(index_name, taskID, request_options)
	if status == 'published'
	return
	end
	sleep(time_before_retry.to_f / 1000)
	end
	end