I've recreated an VM instance that uses the curtin config attached to
this bug.

I can recreate the additional registrations and delete/create of the
dname symlinks to the bcache devices by simply running 'udevadm trigger'
in a loop, and watching, dname directory with inotifywatch (inotifywait
-mr /dev/disk/by-dname).

During a 'udevadm trigger', I can observer 10 DELETE operations
(bcache0..9), and then 10 CREATE operations (bcache0..9).

The symlinks appear almost immediately.

Next, I increased the IO load on the system by spinning up fio against
several of the bcacheN devices;  This increased the time between removal
and creation, but only in the order of milliseconds.

I've not been able to induce a state such that only some of the bcache*
symlinks are missing; so we're still looking for what else is going on
in the system such that we've got pending uevents that need to be
processed.

Do we have any insight into the program using the dname symlinks?

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1812925

Title:
  No OSDs has been initialized in random unit with "No block devices
  detected using current configuration"

To manage notifications about this bug go to:
https://bugs.launchpad.net/charm-ceph-osd/+bug/1812925/+subscriptions

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to