Hi all, Currently, Object watcher just register timeout callback function HandleWatchTimeout on primary OSD. However, If primary OSD and rbd client becomes down at the time (If virtual machine and ceph-osd is on the same physical server) , then the replica OSD will leave the watcher there until the Object is accessed and populate the watcher again. So eventually, if we want to remove the unused image, it will return error saying image is still busy because the watcher is still there. Currently, if we want to find out the object containing watchers, we need to scan all objects so that it is unreasonable to do that. But I think it is reasonable to remove the watcher normally in this case, otherwise client will get wrong information about whether the image is busy. So we can record the watcher information in omap and retrieve the information the cluster restart or something primary OSD is down? Regards Ning Yao -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
