Tom Parrott

Full details from OP on getting this working with suggestions on how to improve the docs: https://discourse.ubuntu.com/t/advertising-ovn-networks-over-bgp/61198/14?u=tomp

@hamistao to be comfortable with this change please can you double check that the /dev/disk/by-id names do not change for block disk device passthroughs. Cheers

> I beleive this is safe to merge. Both entries for the device on `/dev/disk/by-id` are generated from the block node name (and not the device ID), so this could...

@massigori please can you look for any AppArmor denials on the host that is running the container please?

@massigori please can you show the `lxc config show --expanded` output for the container too. I'm wondering if its a problem with the UID/GID range being used that is too...

@massigori please can you also get the output of: ``` lxc exec -- cat /proc/1/uid_map lxc exec -- cat /proc/1/gid_map ```

There's this article that describes the various options around UID/GID mapping: https://ubuntu.com/blog/custom-user-mappings-in-lxd-containers It would be great to get authd to use a range less than 1000000000 as then it'll work...

Thanks! It would be good if it uses the `dqliteDriver.Error` error code too rather than string matching. If possible. Like: https://github.com/lxc/lxd/blob/50b4f4aa27a863998e9ea3079625fbaecb44d5e6/lxd/db/storage_buckets.go#L541-L543

Could I request that we dont panic in the library as that will prevent the app from taking any remedial/notification action.

> I'm happy to put up another PR against `stable-5.21` with the "wrong way" to fix #12567 (simply checking against a literal `"powerflex"` in `lxd/project`) if that would be preferable...