Kohei Tokunaga

Yes. You can use `ctr-remote` command for converting the original image into stargz and pushing it to registries.

We are seeing an issue for using cloud-provided registries(e.g. GCR). so if you are to try it with these registries, could you wait for #81 get merged? This will be...

And thanks a lot for trying it out!

@andrewrynhard #81 got merged so this should work with cloud-provided registries. Can you try it out? You can convert the hyperkube image by `ctr-remote image optimize `.

@itsmurugappan Thanks for reporting this. Does this occur everytime the snapshotter pulls `ghcr.io/itsmurugappan/java-hw-estgz:v1` ?

@luodw Thanks for the question! Though we have graceful shutdown on SIGINT (#26), recovery on abnormal shutdown / support for service restart are in progress (#134). Very welcome for contribution.

@luodw Can you check if the master version (contains the patch #134) fixes this issue?

Currently, you need to re-run containers too. And I agree with that the snapshotter needs to be able to gracefully restart even if there are running containers.

Currently, we need to kill all containers running on that node before restarting this snapshotter and re-deploy these containers after the snapshotter restarts. One of the idea to solve this...

> trickle layout We haven't tried trickle layout but patches are welcome if they improve the performance. > the time to first byte When lazy pulling eStargz, stargz-snapshotter firstly reads...