Currently, if all search engines timeout, the user doesn't have any result to read.
In this scenario, the user usually refresh the page to make sure there is really no result : and sometimes it displays some results.

Technically, this can happen when the HTTPS connections have been closed. On the second try, the connections are already opened, and the results arrived before the global timeout.

See : https://news.ycombinator.com/item?id=14579491#14581065

I see two solutions (most probably there are a lot of more) :

• advise the administrator to increase the global timeout in this case. How ?
• change the searx behavior : if there is no results after the timeout, double the timeout and wait more.