Using CTDB and Samba on Gluster

Mounting gluster volume using Samba alone does not implement High Availability, but CTDB can solve this problem.

1 prepare the environment

I have prepared a gluster environment in advance, created a volume repvol, and this volume has been configured to be mounted using samba client. Please refer to my previous blog post for detailed configuration.

[root@node02 /] # gluster pool listUUID Hostname Statee967c77f-ff39-4927-9645-606294d42751 node01.lab.example.com Connected 98ef6fd9-9fa1-4005-b8ec-5480d6c04f46 localhost Connected [root@node02 /] # gluster volume listrepvol [root@node02 ~] # smbclient-L node01.lab.example.com-U%

Domain= [SAMBA] OS= [Windows 6.1] Server= [Samba 4.4.4]

Sharename Type Comment

-

Print$ Disk Printer Drivers

Gluster-repvol Disk For samba share of volume repvol

IPC$ IPC IPC Service (Samba 4.4.4)

Domain= [SAMBA] OS= [Windows 6.1] Server= [Samba 4.4.4]

Server Comment

--

Workgroup Master

--

[root@node02 mnt] # mount-t cifs-o username=smbuser,password=redhat / / node01.lab.example.com/gluster-repvol / mnt/samba-test/ [root@node02 /] # df-h | grep samba//node01.lab.example.com/gluster-repvol 2.0G 33M 2.0G 2% / mnt/samba-test

As you can see from the above, I have two gluster server here, one by one

192.168.133.144 node01.lab.example.com

192.168.133.149 node02.lab.example.com

When mounting, I use / / node01.lab.example.com/gluster-repvol, so there is a problem. Although my gluster volume repvol here is a 2-copy replication volume, once the node01.lab.example.com fails, although there is still a copy of data on the node02.lab.example.com, it cannot be directly used by client and cannot be switched automatically.

2 install and configure ctdb on two gluster server

2.1 install ctdb using yum

[root@node01 ~] # yum install-y ctdb

Release firewall restrictions for ctdb

[root@node01] # firewall-cmd-- add-port=4379/tcp

2.2 Update volume settings

[root@node01 ~] # gluster volume stop repvol [root@node01 post] # cd / var/lib/glusterd/hooks/1/start/post [root@node01 post] # vi S29CTDBsetup.sh

Before the modification of META= "all"

After the modification of META= "ctdbmeta"

2.3 Update Samba configuration

Edit the / etc/samba/smb.conf file and add clustering=yes under the [global] option

[root@node01 post] # cat / etc/samba/smb.conf... [global] clustering = yes....

2.4start volume, you will find that volume will be automatically mounted to / gluster/lock of server.

[root@node01 post] # gluster volume start repvolvolume start: repvol: success [root@node01 post] # df-h | grep repvolnode01.lab.example.com:/repvol.tcp 2.0G 33M 2.0G 2% / gluster/lock

2.5 add / etc/ctdb/nodes and / etc/ctdb/public_address files

[root@node01 post] # vi / etc/ctdb/nodes 192.168.133.144192.168.133.149

# ip here is the ip of gluster server node

[root@node01 post] # vi / etc/ctdb/public_addresses 192.168.133.100/24 ens33

# here ip is floating ip and ens33 is interface name

2.6 start and verify ctdb service

[root@node01 post] # systemctl start ctdb [root@node01 post] # ctdb pnnPNN:0 [root@node01 post] # ctdb ping-n allresponse from 0 time=0.000049 sec (2 clients) response from 1 time=0.008014 sec (1 clients) [root@node01 post] # ctdb statusNumber of nodes:2pnn:0 192.168.133.144 OK (THIS NODE) pnn:1 192.168.133.149 OKGeneration:2009096591Size:2hash:0 lmaster:0hash:1 lmaster:1Recovery mode:RECOVERY (1) Recovery master:1 [root @ node01 post] # ctdb ipPublic IPs on node 0192.168.133.100 0 [root@node01 post] # ctdb ip info 192.168.133.100Public IPs on node 0192.168.133.100 0

2.7Mount gluster volume using floating ip

[root@node02 post] # mount-t cifs-o username=smbuser,password=redhat / / 192.168.133.100/gluster-repvol / mnt/samba-test/