ThinkingSphinx::ConnectionError - Error connecting to Sphinx via the MySQL protocol. Lost connection to server during query

[madmax]

After upgrading to recent manticore search and going to docker version https://hub.docker.com/r/manticoresearch/manticore/ (6.3.6) i get sometimes error

ThinkingSphinx::ConnectionError

Error connecting to Sphinx via the MySQL protocol. Lost connection to server during query - SELECT *, WEIGHT() as rank FROM `movie_search_0_core`, `movie_search_1_core`, `movie_search_2_core`, `movie_search_3_core` WHERE MATCH('@(titles,year_index) _begin o que podem as palavras _end') AND `sphinx_deleted` = 0 ORDER BY `rank` DESC, `popular` DESC LIMIT 0, 20 OPTION cutoff=0, field_weights=(titles=20, year_index=1), ranker=proximity; SHOW META

Checked server and there is no crash for manticore. It happens very rarely. Any idea how debug this kind of errors?
I know there is Retrying query mechanism but untill but only works for ThinkingSphinx::QueryExecutionError not for ThinkingSphinx::ConnectionError.

Running the same query again works fine. It looks like this has nothing connected to specific query.

[madmax]

ThinkingSphinx::ConnectionError actually make query retry. Increased retry times from 3 to 10. But was wondering how often this retry is used. I subscribe message.thinking_sphinx and logs in sentry.

ActiveSupport::Notifications.subscribe("message.thinking_sphinx") do |event|
  if event.payload[:message] && event.payload[:message].match?(/^Retrying query/)
    Sentry.set_extras(attributes: event.payload)
    Sentry.capture_message("Error Sphinx query")
  end
end

It's normal that this happens so often? For normal mysql connection there is no retry mechanism and I can't see this kind of errors. Also turned on debugging on manticoresearch but there is no information that something is wrong when ost connection to server during query appears.