You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I don't know what you're trying to do, but the point of using a FIFO is that it allows us to send a signal to runc init in a way that can't be missed and in a way that many runc start invocations will be a no-op (effectively, something like a pthread_barrier_t EDIT: actually, pthread_barrier_t doesn't work this way).
Many years ago we used to do this by sending signals, but signals can get lost or coalesced for a variety of reasons and you don't want to send signals to the actual container process (even if they are ignored by default for pid1).
A message-based scheme with a socket would struggle with being able to handle many runc starts AFAICS.
Description
runc init can't send message to runc start if use exec.fifo.
Steps to reproduce the issue
runc init can't send message to runc start if use exec.fifo.
Describe the results you received and expected
runc init can send message to runc start by using exec.sock
What version of runc are you using?
1.12
Host OS information
any kernel
Host kernel information
any
The text was updated successfully, but these errors were encountered: