What bothers me is that I can implement fencing according to your video, but when I unplug the fiber cable of the node's multipath to test, the node whose fiber cable was unplugged did not restart. In this case, the node whose fiber cable was removed needs to be fenced (rebooted).
I see you used /dev/sdb as your device instead of uuid for the device. When I rebooted my nodes the scsi devices /dev/sda & /dev/sdb were swapped and my cluster wouldn't start because fencing was broken.
Hello After restarting cluster, the sdb -d command only shows local server. I am sorry but i cant paste the output Can you figure out what's happening? Thanks
![[RHEL8] Redhat Pacemaker PCS Cluster | Configure HA Cluster- Prerequisite- RHEL8 |CentOS8](http://i.ytimg.com/vi/c1e3ZC4Ppas/mqdefault.jpg)
![[RHEL8] Redhat Pacemaker PCS Cluster | Configure HA Cluster- Prerequisite- RHEL8 |CentOS8](/img/tr.png)
What bothers me is that I can implement fencing according to your video, but when I unplug the fiber cable of the node's multipath to test, the node whose fiber cable was unplugged did not restart. In this case, the node whose fiber cable was removed needs to be fenced (rebooted).
Great Its working fine
Thank you.
Thanks ,this is a good video.
Great Video friend.... it helps me a lot... pl help to configure oracle db on pacemaker cluster with failover with LVM
Hello sir, do we need to create physical volume for both nodes or just 1 node is enough
I see you used /dev/sdb as your device instead of uuid for the device. When I rebooted my nodes the scsi devices /dev/sda & /dev/sdb were swapped and my cluster wouldn't start because fencing was broken.
Is the same procediment with multipath block devices? /dev/mapper/mpatha....
Hi , how to solve this error msg ''fence_sbd: power timeout needs to be greater then sbd message timeout' ?
You didn't mention that how did you set up shared storage?
what software you are using for taking ssh connection to server?
Mobaxterm
Hi. I don’t quite understand. Where is the shared storage in this context ? And why do u need to create fencing stonith 2 times when it should be 1 ?
/Dev/sdb is the shared disk I am using for fence device. I have created only 1 stonith agent and that is enough.
why SBD device in same cluster nodes ?
Hello
After restarting cluster, the sdb -d command only shows local server.
I am sorry but i cant paste the output
Can you figure out what's happening?
Thanks
Please watch the video again and follow exact steps I did. I will not able to comment without the error.
@@tunetolinux4173 Thanks. I think the error is that /dev/sdb is a different disk in each VM. Can it be?
Please paste all the commands from notepad under description, it would be helpful
1. Install fence-agents-sbd package on both nodes.
yum -y install sbd fence-agents-sbd
2. On node01, Initialize the disk and create sbd device.
sbd -d /dev/sdb -4 20 -1 10 create
3. Check status of sbd device.
sbd -d /dev/sdb dump
==Dumping header on disk /dev/sdb
Header version : 2.1
UUID : 3a77e11e-f9ee-4f6c-bbd8-d8570cc1f694
Number of slots : 255
Sector size : 512
Timeout (watchdog) : 10
Timeout (allocate) : 2
Timeout (loop) : 1
Timeout (msgwait) : 20
==Header on disk /dev/sdb is dumped
4. Enable software watchdog
modprobe softdog
echo softdog > /etc/modules-load.d/softdog.conf
5. Edit sbd configuration file and below two information on both the nodes.
# edit sbd config /etc/sysconfig/sbd
SBD_DEVICE="/dev/sdb"
SBD_OPTS="-W"
6. Enable sbd service and restart cluster service.
systemctl enable sbd
pcs cluster stop --all
pcs cluster start --all
7. List sbd devices.
sbd -d /dev/sda list
0 clnode01 clear
1 clnode02 clear
# get sbd status using fence_sbd, and then reboot node02 from node01
fence_sbd --devices=/dev/sdb -n clnode02 -o status
fence_sbd --devices=/dev/sdb -n clnode02 -o reboot
stonith_admin --reboot clnode02
8. Setup the fencing on pcs
pcs stonith create fence-sbd fence_sbd devices="/dev/sdb" power_timeout=20
9. Test fence node with fence agent just created.
pcs stonith fence clnode02
10. Check sbd device information.
pcs stonith sbd status