I am unaware of any way to accomplish having 1 pool with all 3 racks and another pool with only 2 of them. If you could put the same osd in 2 different roots or have a crush rule choose from 2 different roots, then this might work out. To my knowledge neither of these is possible.
What is your reasoning for using min_size of 1? I would highly recommend away from min_size = 1 for anything other than maintenance. Most people that use it believe that it means that writes will return after it has been written to min_size, but that is not the case. Writes happen to every osd in the PG before it acknowledges it is complete. Min_size is merely how many copies of a PG need to be up for writes and reads to still happen. Running with that set to 1 has a higher chance of data corruption than I'm comfortable with. Look through the mailing list archives to see more about this.
Hello,
I'm building a 5 server cluster over three rooms/racks. Each server has 8
960GB SSDs used as bluestore OSDs. Ceph version 12.1.2 is used.
rack1: server1(mon) server2
rack2: server3(mon) server4
rack3: server5(mon)
The crushmap was built this way:
ceph osd crush add-bucket rack1 rack
ceph osd crush add-bucket rack2 rack
ceph osd crush add-bucket rack3 rack
ceph osd crush move rack1 root=default
ceph osd crush move rack2 root=default
ceph osd crush move rack3 root=default
ceph osd crush move server1 rack=rack1
ceph osd crush move server2 rack=rack2
...
rule replicated_rule {
id 0
type replicated
min_size 1
max_size 10
step take default
step chooseleaf firstn 0 type rack
step emit ^^^^
I had to manually change the replicated_rule in the crushmap. Is this change
also possible via cli?
- step chooseleaf firstn 0 type host
+ step chooseleaf firstn 0 type rack
The first created pool has size=3/min_size=1 and every copy is in a different
rack. With this setup the cluster is loosing the capacity of 2 servers because
rack3 has only one server.
I'd like to add a second pool with size=2/min_size=1 and a rule to but copies
only in rack1 and rack2.
Is that possible or should I think about a complete different solution?
Thanks in advance,
Andreas
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
