Use http read total timeout by default (logging errors rather than raising)
Compare changes
Files
19- Andy Schoenen authored
Removes the use_read_total_timeout option on Gitlab::HTTP and use read_total_timeout by default. The option was introduced for HTTP calls to URLs that could be defined by users in order to fix a security issue but have the smallest possible impact. It looks like the timeout implementation is stable so it can become the default option. The only exception where We don't want to use a read total timeout is when streaming potentially large responses.