RoadRunner allocates workers like usual processes (via fork + exec syscalls). While RoadRunner is in the process of creating a worker (connecting to the pipes, TCP, etc.), some part of the worker might freeze the initial handshake. RoadRunner waits pool.allocate_timeout time for handshake to complete and return this error if pool.allocate_timeout exceeded.
How to fix that?
pool.allocate_timeout option. It should be in the form of pool.allocate_timeout: 1s or pool.allocate_timeout: 1h, so you should specify the units of measurement. Also, keep in mind, that 1s for the allocate timeout might be a very small value, try to increase it.xdebug might freeze the worker spawn while improperly configured. See this tutorial: link
pipes, check this options: https://github.com/roadrunner-server/roadrunner/blob/master/.rr.yaml#L75. It's responsible for the initial handshake timeout between RR and PHP process established via sockets or TCP.