Pad memory volumes to 4096 bytes to support 4K storage. This was fixed only partly in 4.3.
Raising priority and severity since this can cause failures when cloning vms with memory volumes.
Fred, anything missing in this bug for including in 4.3.6?
I see you already targeted it. The description of the bug is very vague. Missing: what is the scenario/flow, what is the failure, user impact, reproducible, which storage (only 4k?) QE should be able to verify this.
Without this change, qemu may segfault when copying memory volumes (e.g bug 1649788). This issue may have been fixed in qemu, but in vdsm we should not create images which are not aligned. In the past we aligned images to 512 bytes, but when using storage with sector size of 4096 bytes, images should be aligned to 4096 bytes. This affects only NFS storage since other types of file storage uses fileutils.padToBlockSize() which is already fixed in 4.3.6 (or before). I don't know if it is possible to reproduce this qemu segfault, it depends on qemu version and the contents of the memory volume. To verify the fix, you can check the size of a memory volume created when taking a snapshot with memory. The size must be a multiple of 4096.
Hi Pavel, Please provide a clear reproduction/verification scenario so we can QA_ACK this bug.
verification scenario is: 1) Create a snapshot with memory thus creating a memory volume 2) Check the memory volume size(in Bytes) is a multiple of 4096.
Verified on: vdsm-4.30.29-1.el7ev.x86_64 ovirt-engine-4.3.6.4-0.1.el7.noarch
This bugzilla is included in oVirt 4.3.6 release, published on September 26th 2019. Since the problem described in this bug report should be resolved in oVirt 4.3.6 release, it has been closed with a resolution of CURRENT RELEASE. If the solution does not work for you, please open a new bug report.