This is a multi-part message in MIME format. --------------080509020005000107080304 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit On 16/01/15 19:26, Mario Giammarco wrote:
2015-01-16 12:37 GMT+01:00 Simone Tiraboschi <stirabos@redhat.com <mailto:stirabos@redhat.com>>:
HA capability is provided for other VMs by oVirt engine. But who provide it if the engine itself is on a VM on the host that it's managing? HA for the Engine VM needs to be managed by the hosts and not the Engine itself: so we have ovirt-hosted-engine-ha that ensure HA for the engine VM, the engine cloud than provide HA for other VMs.
I am surprised. I supposed that HA was "self provided" by the cluster like in xenserver. So you tell me that is the engine that checks if servers and vms are on like in cloudstack?
This is just how any VM self-hosted setup would work. The 'engine/management' VM has to have HA managed by something other than the engine itself - otherwise if the engine is down how would it know or be able to restart itself? In VMWare or Xenserver there would have to be a separate system other than that in the engine VM to make sure that the management engine VM is a) running on at least one host on the cluster and b) *cannot* be running on more than one host to avoid screwing its own storage volume (ie heartbeat/fencing). Then this "managed engine" only has to take care of keeping its own VMs up. Logically I cannot see any other way this could possibly work - see "chicken and egg"! Cheers Alex --------------080509020005000107080304 Content-Type: text/html; charset=windows-1252 Content-Transfer-Encoding: 8bit <html> <head> <meta content="text/html; charset=windows-1252" http-equiv="Content-Type"> </head> <body bgcolor="#FFFFFF" text="#000000"> <br> <div class="moz-cite-prefix">On 16/01/15 19:26, Mario Giammarco wrote:<br> </div> <blockquote cite="mid:CABQ2P6GF92-ZuOOZAv3Gu98wg-0VM_JStV98N8j68DixADnCog@mail.gmail.com" type="cite"> <div dir="ltr"><br> <div class="gmail_extra"><br> <div class="gmail_quote">2015-01-16 12:37 GMT+01:00 Simone Tiraboschi <span dir="ltr"><<a moz-do-not-send="true" href="mailto:stirabos@redhat.com" target="_blank">stirabos@redhat.com</a>></span>:<br> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class=""><br> </span>HA capability is provided for other VMs by oVirt engine. But who provide it if the engine itself is on a VM on the host that it's managing?<br> HA for the Engine VM needs to be managed by the hosts and not the Engine itself: so we have ovirt-hosted-engine-ha that ensure HA for the engine VM, the engine cloud than provide HA for other VMs.<br> <br> </blockquote> </div> I am surprised. I supposed that HA was "self provided" by the cluster like in xenserver. So you tell me that is the engine that checks if servers and vms are on like in cloudstack?<br> </div> </div> <br> </blockquote> <br> This is just how any VM self-hosted setup would work. The 'engine/management' VM has to have HA managed by something other than the engine itself - otherwise if the engine is down how would it know or be able to restart itself? In VMWare or Xenserver there would have to be a separate system other than that in the engine VM to make sure that the management engine VM is a) running on at least one host on the cluster and b) <b>cannot</b> be running on more than one host to avoid screwing its own storage volume (ie heartbeat/fencing).<br> <br> Then this "managed engine" only has to take care of keeping its own VMs up. Logically I cannot see any other way this could possibly work - see "chicken and egg"!<br> <br> Cheers<br> <br> Alex<br> <br> <br> </body> </html> --------------080509020005000107080304--