start up a Ceph cluster.
I m interested in working on the Ceph project "Creation of a CephFS Shell
and CLI Tool" and hence had filled the application form for the same.
As part of the initial step "start up a Ceph cluster" i have created a Ceph
cluster. I have used ceph-deploy tool to create Ceph cluster, and a Ceph
client(filesystem). Followed, the documentation in [1][2][3][4]. The setup
has 4 centos 7.4 VMs running on Fedora 27 host. One of the nodes is
admin-node and the others are osd,mon,mgr node. Have used one of these
nodes in the cluster as client node and have created a Ceph file system
mount. Here are the details of the cluster:
[root@vm2 ~]# ceph status
cluster dc1e5542-eb3f-49a7-99fb-bc00f96c91ab
health HEALTH_OK
monmap e1: 1 mons at {vm3=192.168.100.132:6789/0}
election epoch 3, quorum 0 vm3
fsmap e5: 1/1/1 up {0=vm1.localhost.net=up:active}
osdmap e24: 3 osds: 3 up, 3 in
flags sortbitwise,require_jewel_osds
pgmap v54: 66 pgs, 3 pools, 2068 bytes data, 20 objects
103 MB used, 3007 MB / 3110 MB avail
66 active+clean
[root@vm2 ~]#
[root@vm2 ~]# ceph fs ls
name: cephfs, metadata pool: cephfs_metadata, data pools: [cephfs_data ]
[root@vm2 ~]# date
Tue Mar 27 03:02:32 EDT 2018
[root@vm2 ~]# df -h
...
192.168.100.132:6789:/ 3.1G 104M 3.0G 4% /mnt/mycephfs
[root@vm2 ~]#
Initially i had faced some issues with deploy, that was because i had used
fedora host as the admin node, and there was mismatch in the ceph-release
package. I then shifted the admin node to another centos VM, then the
installation was smooth. Here is the link to the install log from the admin
node[5] .
I am currently working on the second task " create and run a simple C
program that uses the libcephfs library in some way.". Will post the patch,
soon.
[1]
http://docs.ceph.com/docs/master/start/
[2]
http://docs.ceph.com/docs/master/start/quick-start-preflight/
[3]
http://docs.ceph.com/docs/master/start/quick-ceph-deploy/
[4]
http://docs.ceph.com/docs/master/start/quick-cephfs/
[5]
https://paste.fedoraproject.org/paste/Sc5KeFJPihIcqq8mBvFFNw
Thanks,
Ranjitha.