Forum
Welcome, Guest
Username: Password: Remember me
This is the optional category header for the Suggestion Box.

TOPIC:

HA-SCSI Slave not become itself a master 9 years 9 months ago #260

  • iptelekom
  • iptelekom's Avatar Topic Author
  • Offline
  • Posts: 7
Hi,
We are month with HA, I hope at the end of the test environment. All as You wrote is working, but I have additional question to You.

Please explain us how is looks like a procedure of corectly shutting down the machine:
- both of them (for example we want move both of them to new location)
- first one machine, next another during working pool (for example, we have to upgrade software, but we have to has also a working pool)

This is what we tried to do:
1. We have master and slave, masters goes down by "Shut down" option and after it slave is rebooted by own, we have got this (full logs are in attachment)

Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 Function fence_host returned status 1
Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 Function fence_host returned status 1
Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 Failed to fence Master: c277f5d2-ce5f-46b6-859d-cf0a39ada1ca. Checking whether FENCE_HA_ONFAIL is enabled
Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 Failed to fence Master: c277f5d2-ce5f-46b6-859d-cf0a39ada1ca. Checking whether FENCE_HA_ONFAIL is enabled
Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 Marking this host as fenced, Rebooting this host now!
Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 Marking this host as fenced, Rebooting this host now!
Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 !!!!!!!!!!!!!!!!!!!!!!!!!!!!! SELF FENCING - REBOOT HERE !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Jul 4 14:46:55 169.0.0.71 ha-lizard: 16824 !!!!!!!!!!!!!!!!!!!!!!!!!!!!! SELF FENCING - REBOOT HERE !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

After rebooting slave notice as that he could find a master of pool and vm-s are disconnected, slvae don't have a network cards...

We have to tunr on the master, reboot slave once again and everything back to working

2. We have master and slave, masters goes down by hardly disconnecting the power supply and after thaht slave become a master, master is turned off - good, but this is hardly shut down.

When we turn on the machine, it become a slave.

3. We move both machine to manual iscsci, how its described in documentation, we want change master to slave and slave to master manualy and turn off the slave (originally the master) - how we should do that correctly

4. When we turned off both of machine, we have to remeber which of them is master, and turned it on first to have pool working, if we don't do this, we are getting info that slave could not find a master and pool is working with one machine.

I will be grateful for any help
Attachments:

Please Log in or Create an account to join the conversation.

HA-SCSI Slave not become itself a master 9 years 9 months ago #261

If you need to shutdown any of the hosts you should Always disable HA-Lizard first. In looking at your slave log it looks like the master was already gone.

Regarding moving your servers. If moving both at the same time:
1- disable HA-Lizard 2- Shutdown ALL VMs 3- Shutdown the hosts

If moving one at a time with no VM downtime - see appendinx C of this document which provides step by step details. In this case it is assumed that the replication link be available.
halizard.com/images/pdf/iscsi-ha_node_cl..._howto_1.4_final.pdf

Please Log in or Create an account to join the conversation.