wido commented on PR #14320: URL: https://github.com/apache/cloudstack/pull/14320#issuecomment-6142890910
> > I like the watchers method the best as its’s as native to RBD as possible. > > yeah, I like it too. I just realized a potential performance issue. Imagine an environment with thousands of volumes, in the worst case, if a KVM host goes down, running rbd status against every volume could take a long time to complete. Wouldn't a single RBD image per host be sufficient and have it watch it for ever? If it stops watching, others can assume it's dead. You could also have it write the timestamp in metadata (omap) instead of writing RADOS objects. These RADOS objects in the Ceph pool will go stale when you remove a host. Yes, small objects, but it becomes garbage. RBD images are easier to spot by other tools. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
