TOPIC:

Hello, currently we backup our VM's via LVM Snapshots, is there a way to configure ha-lizard to use LVM or is it already LVM? Also is there a way too keep some space open to make lvm snapshots?

Second is there a cmdline or some api to get the active master node on ha-lizard via a commandline tool? Currently we parse drbd-overview on xenserver so that we only make backups on the primary.

Also will the secondary be the new master after a reboot of the master or will it wait till the master comes online? Last week we had a tape failure and needed to replace the tape drive so we manually shutdown our previous master.

And is there a easy way to keep the Pool up to date without destroying ha-lizard?

Thanks,
Christian Schmitt

Hi Christian,
Assuming you are using ha-lizard + iscsi-ha in a 2-node pool, the storage is already using LVM. Actually, XenServer would have added the LVM layer when introducing the iscsi-ha storage to xenserver for the first time. If you want to carve out some of the storage for another purpose, I suppose you could partition the disk prior to installing our solution and then assign only one partition to DRBD, leaving the other free to do what you wish.

Regarding command line tools. Again, assuming you have installed both of our applications (ha-lizard and iscsi-ha), you have 2 command line tools available:
iscsi-cfg - this will give you what you are looking for. "iscsi-cfg status" will provide you with DRBD live status and tell you which host is master. Also, iscsi-cfg provides the ability to move the storage (live) between hosts, allowing for host reboots without losing access to storage.
ha-cfg - this interface provides full control over HA functionality

Both command line tools are extensively documented. Please refer to the documentation for more.

In the event of a master reboot the secondary will not promote itself to master. The correct procedure for rebooting a master host is as follows:
- disable HA via ha-cfg CLI
- put iscsi-ha into manual mode
- expose storage on the slave via iscsi-cfg CLI

you can now safely reboot your master and your VMs will continue to run on the slave. If you don't perform these steps your VMs will likely lose connectivity to the storage while the master is down.

Regarding your last question about keeping a pool up to date. As of XenServer 6.5, nearly all xenserver updates have no affect on HA-Lizard. Generally only major upgrades (say version 6.2 to 6.5) will affect you. We do plan on changing our SW packaging in the near future to properly prepared RPMs that can be installed into XenServer.

Wow great! Just one other question, what happens when the master get's shutdown in a failure event? Will the slave then be promoted? Like power outage, hardware failure, etc.?

If the master fails, the slave will promote itself to master and start any VMs that were running on the failed master. To properly test this you should crash your master (not reboot)