mgr/cephadm: fix and improve osd draining#39536
mgr/cephadm: fix and improve osd draining#39536sebastian-philipp merged 6 commits intoceph:masterfrom
Conversation
When adding an osd daemon explicitly, there is no created timestamp for the spec, and we should never not apply it. Fixes: b129c13 Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
If we are replacing an OSD, we should mark it out and then back in again when a new device shows up. However, if we are going to destroy an OSD, we should just weight it to 0 in crush, so that data doesn't move again once the OSD is purged. Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
Otherwise it doesn't work!
Drop the fullname property: it is always "osd.{self.osd_id}".
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>
|
oops. this broke the upgrade test: |
I don't think this is related to my pull request. The process_removal_queue() method is called at the top of the serve loop and it unconditionally saves the list. (also, that run was a pass) |
|
@sebastian-philipp @liewegas Whether this pr will backport to octopus? In the issue #49339, the backport is described as 'pacific,octopus'. |
|
do you want to do the backport? That's mainly a |
Uh oh!
There was an error while loading. Please reload this page.