Ayato Tokubi
Ayato Tokubi
@sj98ta I just want to rule out the possibility other application controls systemd. If you are running something on the joining node, I want them to stop and try again....
> Again, it does not happen every time, and this last occurrence, I had it looping the join/reset procession for over an hour before it hit (which made for a...
@sj98ta Did some tests, and found StartTransientUnit should work even if the slice is not created. This is what I ran. ```bash sudo gdbus call --system \ --dest org.freedesktop.systemd1 \...
@sj98ta ``` ❯ systemctl --version systemd 256 (256.5-2ubuntu3.1) +PAM +AUDIT +SELINUX +APPARMOR +IMA +SMACK +SECCOMP +GCRYPT -GNUTLS +OPENSSL +ACL +BLKID +CURL +ELFUTILS +FIDO2 +IDN2 -IDN +IPTC +KMOD +LIBCRYPTSETUP +LIBCRYPTSETUP_PLUGINS +LIBFDISK...
@sonnysasaka > This option should ensure that image storage is never wiped, regardless of the circumstances kubelet also removes container images via CRI, should it also be prevented? https://kubernetes.io/docs/concepts/architecture/garbage-collection/#containers-images In...
I think just stopping deprecation of `internal_repair` is enough. For failed repair wipe, it's actually controlled by `internal_repair`. https://github.com/cri-o/cri-o/blob/9e4d86d82370ad44c14649f28ebdd8f94aaa28ca/internal/lib/container_server.go#L107-L129 Is there any other scenarios?
sorry I got `internal_repair` mixed up with `internal_wipe`. > Can I also suggest, that if we decide to undeprecate internal_wipe, we also use this flag to control the wipe inside...
@hasan4791 iiuc wipe is inherently not for optimizing the images, but fixing inconsistency. I don't think it's necessary. also GC shouldn't be blocked by this (https://github.com/cri-o/cri-o/issues/9242#issuecomment-2941183209)
@sonnysasaka I wonder if we should prevent wiping when users explicitly run `crio wipe`. Considering the usecase, it might be enough to prevent just automatic wipes.