Normal ZFS bring up at boot is considered as an early boot task by systemd, so LXD won't start before that. The only problem is when your pool isn't mounted by systemd.
In this case, you'd have to manually change the systemd unit to depend on some additional job that you write. Or possibly much simpler, just disable the LXD systemd units and start them manually. The current fallback mechanism is there for people who "upgrade" from plain directory storage over to zfs backed storage, trying to change that logic is extremely likely to cause regressions. Moving forward, we'll have support for multiple storage pools, each with their own state, which will then allow for tracking one pool going down or being unavailable without having this fallback mechanism kick-in. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1642072 Title: lxd uses zfs for images, not containers To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/lxd/+bug/1642072/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
