Hi Volker, Yes, what is described online as instance HA is working in StarlingX. I tested on a cluster with dedicated storage. I added storage extra spec to a flavor, like this: $ nova flavor-key ${flavour_name} set aggregate_instance_extra_specs:storage=remote And I added all the computes to the existing host aggregate named remote_storage_hosts (it already has extra spec storage=remote). Then I launched an instance. And I abruptly stopped the host it was running on. After a few seconds the controller recognized the compute was not responding and the instance was rebuilt on the other compute. === If you still have a question for load-balancing, I recommend a new message to starlingx discuss list with instance load-balancing in the subject line. M On 2018-11-05 11:45 a.m., Michel Thebeau wrote:
Hi Volker,
For context, documents describe Instance HA as "respawn the VM on another compute when the compute the VM was running on fails". But you wrote "load-balancer", from what I can tell these are different subjects. Do you have a document I can refer to?
===
In so far as I am aware, if you kick a compute node over, the VMs that were running there will restart on another compute node in the cluster (assuming another compute can run them). I actually haven't tested this on startlingx though. I'll intend to do that today, now that you've asked about it.
I know nothing about load-balancing.
M
On 2018-11-02 10:15 a.m., von Hoesslin, Volker wrote:
Hello, has starlingX IHA support? I didnt find it? In generally I didn't find HA load-balancer features? What about Octavia https://wiki.openstack.org/wiki/Octavia ?
Greez & thx, Volker...
_______________________________________________ Starlingx-discuss mailing list Starlingx-discuss@lists.starlingx.io http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss
_______________________________________________ Starlingx-discuss mailing list Starlingx-discuss@lists.starlingx.io http://lists.starlingx.io/cgi-bin/mailman/listinfo/starlingx-discuss