some ceph general questions about osd and pg
Hello togehter we want to implement a 3 nodes ceph cluster wirh nautilus. i already tested some ceph installations in our test enviroment and i have some generall questions. end of this month we will have three physical server with 256gb ram and two cpus and nerly 48 x 6tb disks. iam a little bit confused to calculate the pgs with the pgcalc on the ceph side. in the field osds what exactly meands that? the 3 physicaly osd nodes or the disks that will be used for the osds? what can you recomment us? we want later connect the ceph with rados and openstack and lvm thx in advance hfreidhof
Hi Harald, OSD count means the number of disks you are going to allocate to ceph, you can change the whole column by clicking on the "OSD #" at the top of the table. And there's some predefined recommendation for various use cases named: "Ceph Use Case Selector:" you can find it on the same page. I'm not sure if it's really a best practice or not but you may have a look at "All-in-one" use case because you are going to use openstack and rados and lvm. Regards, Khodayar On Sat, Apr 18, 2020 at 9:19 PM <harald.freidhof@gmail.com> wrote:
Hello togehter
we want to implement a 3 nodes ceph cluster wirh nautilus. i already tested some ceph installations in our test enviroment and i have some generall questions. end of this month we will have three physical server with 256gb ram and two cpus and nerly 48 x 6tb disks. iam a little bit confused to calculate the pgs with the pgcalc on the ceph side. in the field osds what exactly meands that? the 3 physicaly osd nodes or the disks that will be used for the osds? what can you recomment us? we want later connect the ceph with rados and openstack and lvm
thx in advance hfreidhof _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io
Hello Khodayar thx fot the fast answer, that helped a lot. only last question: we have two datacenter, a small one with usv(apc) and a big one with a power generator. i want to move two osd nodes to the big datacenter with power generator and one node to the smal datacanter with usv(apc) the reasona are we wanna do this for maintainance things and maybe power or network failures and HA my question is what happens if we have a power problem in the big datacenter, will the node1 in the small datacenter still be a full replicated node and accessible to all client nodes? or will the storage go down? thx in advance, hfreidhof
It's not that simple, you will have to do a lot of edit in the CRUSH map if you want to keep one node in a 3 node cluster as hot backup. IMHO it's not recommended at all to go for such an option. Ceph distributes the data over all the PGs in a pool. If you want redundancy offsite you can have two separated clusters and replicate them with each other. I hope that I made it clear :)
participants (3)
-
doustar＠rayanexon.ir
-
harald.freidhof＠gmail.com
-
Khodayar Doustar