1:45 p.m.
Hello,
just for info... I decided to try this approach and it worked well. I
reinstalled old CentOS7 glusterfs node to CentOS8, enabled ovirt44
repos, installed the same version of glusterfs, mounted bricks, restored
glusterfs and volume (/var/lib/glusterd) configuration. Gluster started
healing and it end soon as brick was offline only for around 1 hour.
Then I add this server as oVirt host and enabled glusterfs service in
cluster.
So far everything looks good (except for probably the unrelated problem
I'll describe in separate mail)
Cheers,
Jiri
On 5/19/21 9:18 PM, Jiří Sléžka wrote:
> Hello,
>
> I'm in progress of moving my oVirt HCI cluster to production. It was a
> bit complicated process because I had two new servers to install oVirt
> from scratch in the lab and three old servers with standalone kvm
> hypervisors in production. One of them (installed with CentOS7) was good
> enough to join HCI cluster. All three old kvm servers had production vms
> running on them.
>
> New cluster was installed with CentOS8 and oVirt4.4, I started with
> single node, then expanded it to two hosts with one another which acts
> as arbiter (just for stability in lab envirnment).
>
> After some testing and tuning I moved this two node cluster (without
> arbiter node) to my server housing. Then I prepared gluster brick on
> that one old server which I want reuse and join it to gluster storage so
> now it is replica 3 - two nodes are CentOS8 based and acts as oVirt
> hosts, one is CentOS7 and acts also as standalone kvm hypervisor.
>
> Then I migrated all vms from standalone kvm hypervisors to oVirt, then
> switched off two oldest hosts.
>
> Now I would like to reinstall CentOS7 node to oVirt host. My plan is to
> keep gluster brick fs and backup configuration and restore it after
> reinstall to CentOS8 as mentioned in
> https://mjanja.ch/2018/08/migrate-glusterfs-to-a-new-operating-system/.
> I want to speed up gluster healing. Does it make sense?
>
> Now the questin: Brick fs was created on CentOS7 and don't have new xfs
> features FINOBT,SPARSE_INODES,REFLINK. Could it be problem? Will be
> better to recreate fs in CentOS8 and then do full heal? I am afraid that
> it will take a long time and make big load to production hosts.
>
> TL;DR: Does gluster/oVirt4.4 make use of FINOBT,SPARSE_INODES,REFLINK
> xfs features?
>
>
> btw. oVirt is great product and HCI looks really functional and usable
> (with 10GE network and SSDsof course). Big thanks to developers and
> community!
>
> Cheers,
>
> Jiri
>
>
>
> _______________________________________________
> Users mailing list -- users(a)ovirt.org
> To unsubscribe send an email to users-leave(a)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/PHPCYOGYL4A...
>