Description of problem: Support for Gluster 4k storage is disabled by default due to issues in qemu and qemu-img, and required testing with gluster 4k storage. Qemu supports now 4k storage since version 2.12.0-33.el7_7.3 (see bug 1745443). Currently users have to install a drop-in configuration file on all hosts to enable gluster 4k support: $ cat /etc/vdsm/vdsm.conf.d/gluster.conf [gluster] enable_4k_storage = true Required changes: - Require qemu-kvm-rhev >= 2.12.0-33.el7_7.3 - Enable gluster:enable_4k_storage when building for RHEL 7 For other platforms (RHEL 8.1, CentOS, Fedora) we will have to wait until fixed qemu version is available.
We cannot enable gluster 4k support before bug 1751722 is fixed, since the code detecting block size on gluster crash gluster fuse mount.
We do not depend on bug 1751722 now since vdsm changed the way block size is detected, so it is not affected by the gluster bug.
Nir, can you please provide a simple way to verify this bug?
As Sas stated here[1] these bug should not be merged to 4.3.7 as we will only be able to test this properly in 4.3.8. Please retarget this bug to 4.3.8. Sas words: " RHHI-V QE team will need more time to run all the regression and this qualification is targeted only RHV 4.3.8. We will start qualification with RHV 4.3.7 with few resource, and this testing will spill over beyond RHV 4.3.7. " So, 4KN feature qualification is NO_ACK wrt RHV 4.3.7 [1] https://bugzilla.redhat.com/show_bug.cgi?id=1765912#c10
We still missing patch enabling 4k storage support for gluster, moving back to POST.
So Nir verification for this bug should be the following, right? 1) check that by default: $ cat /etc/vdsm/vdsm.conf.d/gluster.conf [gluster] enable_4k_storage = true 2) Also check Required changes: - Require qemu-kvm-rhev >= 2.12.0-33.el7_7.3
INFO: Bug status wasn't changed from MODIFIED to ON_QA due to the following reason: [Open patch attached] For more info please contact: infra
(In reply to Avihai from comment #6) > So Nir verification for this bug should be the following, right? > > 1) check that by default: > > $ cat /etc/vdsm/vdsm.conf.d/gluster.conf > [gluster] > enable_4k_storage = true No, this is manual configuration that should not be needed in 4.3.8. To check that 4k is enabled for gluster by default, run: vdsm-client Host getCapabilities And check the output: "supported_block_size": { "FCP": [ 512 ], "GLUSTERFS": [ 0, 512, 4096 ], "ISCSI": [ 512 ], "LOCALFS": [ 0, 512, 4096 ], "NFS": [ 512 ], "POSIXFS": [ 512 ] }, However this will work only after merging this patch, which is still pending: https://gerrit.ovirt.org/c/105398/ > 2) Also check Required changes: > - Require qemu-kvm-rhev >= 2.12.0-33.el7_7.3 Yes.
Nir, I see that all the patches are merged. Why is this bug still on MODIFIED state ? Was the build not happened yet ?
SATHEESARAN, yes, this is included in the last build few weeks ago and should be ready for testing.
Verified ovirt-engine-4.3.8.2-0.1.master.el7.noarch vdsm-4.30.40-1.el7ev.x86_64 Ran the vdsm-client Host getCapabilities and the output was : "supported_block_size": { "FCP": [ 512 ], "ISCSI": [ 512 ], "POSIXFS": [ 512 ], "GLUSTERFS": [ 0, 512, 4096 ], "LOCALFS": [ 0, 512, 4096 ], "NFS": [ 512 ] }, Also checked the required changes: qemu-kvm-rhev-2.12.0-33.el7_7.8.x86_64
This bugzilla is included in oVirt 4.3.8 release, published on January 27th 2020. Since the problem described in this bug report should be resolved in oVirt 4.3.8 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.