Calibrating long cron hook invocations is a challenge; what if, for example, one overloads the task list of the Search hook and it collides with itself on a second cron call ? Having a semaphore protect those instances (and report on hung semaphores if a certain timeout has been reached) would make the multithreading feature more useful.

Comments

vietcoop’s picture

Assigned: Unassigned » vietcoop
MisterSpeed’s picture

Assigned: vietcoop » Unassigned
Status: Active » Needs work

We have a partial solution up (we skip hooks in process), but will get something better down the 2.x line by introducing better hook hangs recoverability