Feature #58368
openInstance from volume snapshot trigger Slow Ops in ceph cluster
0%
Description
We are using openstack ussuri with ceph octopus 15.2.15.
We are using ubuntu 18.04.5 LTS with kernel 4.15.0-122-generic.
We have template volume for create another instances, we use snapshot to duplicate volume and then we create some volumes from that snapshot volume. Sometimes instance created by volume snapshot make the ceph a slow ops.
Steps for reproduce it:
1. Create snapshot from volume template
2. Create some volumes from snapshot in step 1
3, Create some instances with volume in step 2
4. Sometime one of them make a ceph slow ops
We try to restart OSD but slow is moved to another OSD, After that we found out there is pgid suspect slow ops is always same. The pgid is always moved to another OSD after we restarted OSD.
Files
No data to display