[Gluster-users] Gluster 4.1.6 slow

[Prof. Dr. Michael Schefczyk]

Dear All,

For years, I am using virtualization at home on two sets of two hosts running current Centos 7, KVM and Gluster. I am used to pull KVM/libvirt/qemu and Gluster from the ovirt repository, as the KVM/libvirt/qemu versions are more fresh then elsewhere. My VM files are large qcow2-files as I prefer to do a live backup every other day. My Gluster volumes are replica 2 (yes, I do know the downsides, but I did not have issues with that) and since some time sharding (block-size 512MB).

The ovirt repo does include Gluster 3.12. I was (an am) considering to move to Gluster 4.1. Following the upgrade procedure in the documentation, I did upgrade one two-node cluster. That is my smaller cluster running SATA SSD on both nodes and relatively smaller machines. I did not notice any performance hit. After that, I did try to upgrade my second cluster. There, I noticed an unbearable performance hit. The server has one node with SATA SSD and one node with fast SAS HDD. And it runs more machines among them three large Windows VM which see the biggest performance hit.

As experimenting and moving back in a live setting takes multiple hours: Does anyone can point me to potential causes? For the meantime, I did move to NFS storage which does work well. Without better ideas, I will reinstall Gluster 3.12 and see if it does perform as before.

Thanks & Regards,

Michael
____________________________________________________________
Technische Universität Dresden
Fakultät Wirtschaftswissenschaften
Lehrstuhl für Entrepreneurship und Innovation
Prof. Dr. Michael Schefczyk
D-01062 Dresden

Fon: +49-3 51-4 63-3 68 81
Fax: +49-3 51-4 63-3 68 83
www.gruenderlehrstuhl.de