Description of problem: Creating a snapshot out-of-band (e.g. rbd snap create image@snap) on an image with active IO might result in all IO hanging within the QEMU process. This only occurs when the RBD cache is enabled. Version-Release number of selected component (if applicable): 1.3.2 How reproducible: Requires that the RBD cache have pending writeback IO when a snapshot is created. Steps to Reproduce: 1. Start a write-intensive operation within a VM 2. Create a snapshot of the RBD image Actual results: QEMU IO will hang Expected results: QEMU IO continues without issue Additional info:
This issue only affects 1.3.x -- it doesn't affect 2.0. Resetting the flags to account for the change.
Fixed in v0.94.7 upstream. We'll take this BZ as part of the rebase (bz 1335269).
Verified in ceph version: ceph version 0.94.9-1.el7cp Steps followed: 1. enable rbd caching. 2. attach a RBD image to a KVM instance. 3. start IO on the RBD image from the VM. 4. Take snapshots of the RBD image. 5. clone a snapshot. No imapct to IO. Moving this to Verified. Thanks, Tejas
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHSA-2016-1972.html