[Openstack-operators] Problems (simple ones) at scale... Invisible VMs.

Kevin Bringard (kevinbri) kevinbri at cisco.com
Thu May 19 13:49:35 UTC 2016

Given that there’s already the ability to start listing from a specific VM/UUID, it seems like we should add some pagination functionality in there. Maybe we need to add an “—all” flag or something which is non-admin and returns all VMs for the specific tenant. Then if someone really does want all 10,000 VMs they’re running in a tenant they can get them, and still honor osapi_max_limit (per subsequent paged call).

Alternatively (or perhaps in addition to?) if there are more results than the osapi_max_limit, or some other limit set, we should tell people “showing 1000 of 10,000 results, use –marker <last uuid> to show the next 1000” or something.

Limits are something people tend to be OK with, but having to dig into un/under-documented issues is what makes them straight up angry. If there’s something we can do to help alleviate that it seems worth investigating.

On 5/18/16, 5:26 PM, "James Downs" <egon at egon.cc> wrote:

>On Wed, May 18, 2016 at 04:37:42PM -0600, David Medberry wrote:
>> It seems to bypass it... or I'm running into a LOWER limit (undocumented).
>> So help on  limit -1 says it is still limited by osapi_max_limit
>You're looking for:
>--marker <marker>             The last server UUID of the previous page;
>                                displays list of servers after "marker".
>This is much faster than increasing the size of results, at least in sufficiently
>large environments.
>OpenStack-operators mailing list
>OpenStack-operators at lists.openstack.org

More information about the OpenStack-operators mailing list