Hello, you mean this is the reason why the transient queues are not HA? (And thus not quorum when using quorums).

This is still a problem to me, when you lose a rabbit from the cluster. The probability to lose messages is very high. So the probability to have an instance in error is also very high (as a public cloud provider, our api is used a lot, there is a lot of creation/deletion of instances).

Enabling HA for transient queues is helping in that situation.


Le 16 juin 2023 16:06:49 GMT+02:00, "Nguyễn Hữu Khôi" <nguyenhuukhoinw@gmail.com> a écrit :
Hello.
I think that is the reason why people don't use all quorum queue.

https://bugs.launchpad.net/openstack-ansible/+bug/1607830

Nguyen Huu Khoi


On Thu, Jun 15, 2023 at 4:55 PM Arnaud Morin <arnaud.morin@gmail.com> wrote:
Hey team,

I'd like to understand the stakes behing the update_resources_interval
parameter ([1])

We decided on our side to increase this value from 60sec to 600sec (see
[2]).

What I understand is that is will "delay" the update of metrics on nova
side.
I mostly think that these metrics are used by filter scheduler to select
the best host when scheduling.

Is there anything else it can affect?

Cheers,
Arnaud.


[1] https://docs.openstack.org/nova/latest/configuration/config.html#DEFAULT.update_resources_interval
[2] https://review.opendev.org/c/openstack/large-scale/+/886166