No, is not a replica gluster, is just one brick one volume, one single server storage.

This is what I get:

# gluster volume set data group virt
volume set: failed: Cannot set cluster.shd-max-threads for a non-replicate volume.

Since I don't have a replica volume I cannot optimize it for oVirt?

So, it's better to use an iScsi or NFS for a single storage server?

Thanks

José

De: "Strahil Nikolov" <hunter86_bg@yahoo.com>
Para: suporte@logicworks.pt
Cc: "users" <users@ovirt.org>
Enviadas: Terça-feira, 27 De Outubro de 2020 19:20:21
Assunto: Re: [ovirt-users] Improve glusterfs performance

I was left with the impression that you got 2 bricks (2 raids) on a single server.

As I understand , you cannot get a 'replica 3 arbiter 1' or a 'replica 3' volumes , right ?


When you create your volume , you need to use the UI's dropdown "Optimize for virt" or the command 'gluster volume set <VOLUME> group virt' in order to implement the optimal settings for Virtualization.Of course , some tunings can be made like how many I/O threads to be used and some caching - but even the defaults are OK.

Have you checked anything from the list I send you ?


Best Regards,
Strahil Nikolov






В вторник, 27 октомври 2020 г., 19:31:15 Гринуич+2, suporte@logicworks.pt <suporte@logicworks.pt> написа:





Well, each volume has only one brick. Each volume has one dedicated server. Each server storage has the disks in raid10.
I made the gluster volume like this
gluster volume create data transport tcp gfs.domain.com:/home/brick1
And than create a data domain in ovirt: gfs.domain.com:/data

is this correct?

Thanks

José

________________________________
De: "Strahil Nikolov" <hunter86_bg@yahoo.com>
Para: suporte@logicworks.pt
Cc: "users" <users@ovirt.org>
Enviadas: Terça-feira, 27 De Outubro de 2020 17:09:26
Assunto: Re: [ovirt-users] Improve glusterfs performance

Why don't you use the devices of the 2 bricks in a single striped LV or raid0 md device ?

Distributed volumes spread the files among the bricks and the performance is limited to the brick's device speed.

Best Regards,
Strahil Nikolov






В вторник, 27 октомври 2020 г., 18:09:47 Гринуич+2, <suporte@logicworks.pt> написа:





Yes, the VM's disks are on the distributed volume

They were created like this:
gluster volume create data transport tcp gfs2.domain.com:/home/brick1


________________________________
De: "Strahil Nikolov" <hunter86_bg@yahoo.com>
Para: suporte@logicworks.pt
Cc: "users" <users@ovirt.org>
Enviadas: Terça-feira, 27 De Outubro de 2020 15:53:57
Assunto: Re: [ovirt-users] Improve glusterfs performance

Are the VM's disks located on the distributed Volume ?

Best Regards,
Strahil Nikolov






В вторник, 27 октомври 2020 г., 17:21:49 Гринуич+2, suporte@logicworks.pt <suporte@logicworks.pt> написа:





The engine version is 4.3.9.4-1.el7
I have 2 simple glusterfs, a volume with one brick gfs1 with an old version of gluster (3.7.6) and a volume with one brick gfs2 with version 6.8
2 hosts, node3 is up to date, node 2 is not up to date:
RHEL - 7 – 7.1908.0.el7.centos
CentOS Linux 7 (Core)
3.10.0 – 1062.18.1.el7.x86_64
2.12.0 – 33.1.el7_7.4
libvirt-4.5.0-23.el7_7.6
vdsm-4.30.43-1.el7
0.14.0 – 7.el7
glusterfs-6.8-1.el7
librbd1-10.2.5-4.el7
openvswitch-2.11.0-4.el7

gfs1 has 3 nic's gigabit in mode 802.3ad
gfs2 has 4 nic's gigabit in mode 802.3ad

A VM performes better in node3/gfs1:

# dd if=/dev/zero of=test1.img bs=1G count=1 oflag=dsync  1+0 registos dentro 1+0 registos fora 1073741824 bytes (1,1 GB) copiados, 12,3365 s, 87,0 MB/s
in node2/gfs2:

dd if=/dev/zero of=test1.img bs=1G count=1 oflag=dsync  1+0 records in 1+0 records out 1073741824 bytes (1.1 GB, 1.0 GiB) copied, 51.0159 s, 21.0 MB/s

For windows machine is the same. It helps if you have windows drives up to date.

________________________________
De: "Strahil Nikolov" <hunter86_bg@yahoo.com>
Para: suporte@logicworks.pt
Cc: "users" <users@ovirt.org>
Enviadas: Domingo, 25 De Outubro de 2020 19:58:12
Assunto: Re: [ovirt-users] Improve glusterfs performance

It seems that your e-mail went to the spam.

I would start with isolating the issue ?
1. Is this a VM specific issue or a more-wide issue.
- Can you move another VM on that storage domain and verify performance ?
- Can you create/migrate same OS-type VM and check performance?
- What about running a VM with different version of Windows or even better -> Linux ?


Also I would check the systems from bottom up.
Check the following:
- Are you using Latest Firmware/OS updates for the Hosts and Engine ?
- What is your cmdline :
cat /proc/cmdline
- Tuned profile
- Are you using a HW controller for the Gluster bricks ? Health of controller and disks ?
- Is the FS aligned properly to the HW controller (stripe unit and stripe width )? More details on https://access.redhat.com/documentation/en-us/red_hat_gluster_storage/3.5/html-single/administration_guide/index#Hardware_RAID

- FS Fragmentation ? Is your FS full ?

- Gluster volume status ? Client count ? (for example: gluster volume status all client-list)
- Gluster version, cluster.op-version and cluster.max-op-version ?

- If your network is slower than the bricks' backend (disks faster than network) , you can set cluster.choose-local to 'yes'

- Any errors and warnings in the gluster logs ?


Best Regards,
Strahil Nikolov





В четвъртък, 22 октомври 2020 г., 13:59:04 Гринуич+3, <suporte@logicworks.pt> написа:





Hello,

For example, a window machine runs to slow, usually the disk is allways in 100%
The group virt settings is this?:
performance.quick-read=off
performance.read-ahead=off
performance.io-cache=off
performance.low-prio-threads=32
network.remote-dio=enable
features.shard=on
user.cifs=off
client.event-threads=4
server.event-threads=4
performance.client-io-threads=on


________________________________
De: "Strahil Nikolov" <hunter86_bg@yahoo.com>
Para: "users" <users@ovirt.org>, suporte@logicworks.pt
Enviadas: Quarta-feira, 21 De Outubro de 2020 19:22:14
Assunto: Re: [ovirt-users] Improve glusterfs performance

Usually, oVirt uses the 'virt' group of settings.

What are you symptoms ?

Best Regards,
Strahil Nikolov






В сряда, 21 октомври 2020 г., 16:44:50 Гринуич+3, suporte@logicworks.pt <suporte@logicworks.pt> написа:





Hello,

Can anyone help me in how can I improve the performance of glusterfs to work with oVirt?

Thanks

--
________________________________
Jose Ferradeira
http://www.logicworks.pt
_______________________________________________
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-leave@ovirt.org
Privacy Statement: https://www.ovirt.org/privacy-policy.html
oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/
List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/KPU4QFTMUMMFGUA4PYG6624KLSHVLNX4/