Closed sboeuf closed 7 years ago
This PR is blocked by https://github.com/clearcontainers/agent/pull/158 and https://github.com/clearcontainers/agent/pull/160 to be merged in order to behave properly.
@sameo Now everything is passing! I'd like you to give this a second review since I have added patches to solve the issue introduced by the fact we're not waiting for the end of the VM anymore.
@amshinde @sameo @sboeuf the only worry I have with this change is disk cache flushes when using virtio-blk. @amshinde do we do a flush in the VM prior to shutdown? It would be prudent to ensure we do flush before we tore down the VM.
@sboeuf thanks for https://github.com/clearcontainers/agent/pull/169
LGTM
LGTM
This PR first improves the way the tear down of the pod is performed from the agent. Until now, nothing was done and the VM was stopped without tearing down properly the things previously set up inside the VM. Then, it modifies the QEMU implementation for the
stopPod()
method, by not waiting for the VM to exit. Instead, it send a SHUTDOWN through QMP but it does not wait for the VM since there is really nothing it can do.