Ceph Cluster slowness in production
Hello team, I have configured a ceph cluster running on ubuntu 20.04 , with 16 HDD for data and 4 SSD for journaling on each host. the total cluster size is 150 TB usable with a replica of 3. The cluster is in production , integrated with openstack with different pools: Volumes, vms, cinder and cinder-backup. The problem I have is this : For linux servers there is no slowness , but on windows servers is worse , they are very very slow, some of those servers have been migrated from vmware to openstack using coriolis , the network interfaces and switches are 10Gbps. Kindly advise the way I can speed up my cluster to avoid the slowness. Best regards Michel
Hi, first thing coming to mind is to set hw_scsi_model=virtio-scsi and hw_disk_bus=scsi [0]. Or did you already do that? [0] https://docs.ceph.com/en/reef/rbd/rbd-openstack/#image-properties Zitat von Michel Niyoyita <micou12@gmail.com>:
Hello team,
I have configured a ceph cluster running on ubuntu 20.04 , with 16 HDD for data and 4 SSD for journaling on each host. the total cluster size is 150 TB usable with a replica of 3. The cluster is in production , integrated with openstack with different pools: Volumes, vms, cinder and cinder-backup. The problem I have is this : For linux servers there is no slowness , but on windows servers is worse , they are very very slow, some of those servers have been migrated from vmware to openstack using coriolis , the network interfaces and switches are 10Gbps.
Kindly advise the way I can speed up my cluster to avoid the slowness.
Best regards
Michel _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
Hello, I'm not an openstack user myself, but did notice when I was testing with proxmox if I set AIO to native vs io_uring it performs faster in Windows. We don't use Windows normally, so this was just a test environment and don't know if Openstack has something similar. Just my 2 cents for what little it's worth. Cheers On Thu, Dec 12, 2024, 20:28 Michel Niyoyita <micou12@gmail.com> wrote:
Hello team,
I have configured a ceph cluster running on ubuntu 20.04 , with 16 HDD for data and 4 SSD for journaling on each host. the total cluster size is 150 TB usable with a replica of 3. The cluster is in production , integrated with openstack with different pools: Volumes, vms, cinder and cinder-backup. The problem I have is this : For linux servers there is no slowness , but on windows servers is worse , they are very very slow, some of those servers have been migrated from vmware to openstack using coriolis , the network interfaces and switches are 10Gbps.
Kindly advise the way I can speed up my cluster to avoid the slowness.
Best regards
Michel _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
participants (3)
-
Curt
-
Eugen Block
-
Michel Niyoyita