nova_server_resize_poll_interval=5.0 causes intermittent failures with NovaServers.boot_server_from_volume_and_resize

[Nobuto Murata]

NovaServers.boot_server_from_volume_and_resize test case fails intermittently with the following connection error when polling resize status:

ITER: 10 END: Error GetResourceFailure: Failed to get the resource <Server: s_rally_d7e83197_5mMdRhmh>: Unable to establish connection to https://nova.fqdn:8774/v2.1/servers/a1976740-eee9-4582-bed1-b0421f418c41: ('Connection aborted.', RemoteDisconnected('Remote end closed connection without response',))

After looking into the detail, it looks like the default value of nova_server_resize_poll_interval as 5 seconds can hit into a corner case where OpenStack API services are behind Apache2 wsgi and HTTP Keep-alive timeout is 5 seconds as the default value of Apachd2.
https://github.com/openstack/rally-openstack/blob/ec1f31a685bdcbe10212701f8fd0c35598d4745f/rally_openstack/common/cfg/nova.py#L240-L243

We can mitigate the issue by having nova_server_resize_poll_interval explicitly not to use 5 seconds or bump HTTP Keep-alive timeout on the server side. But it would be nice to have nova_server_resize_poll_interval with something other than 5 out of the box.

ref:
https://github.com/psf/requests/issues/4664
https://bugs.python.org/issue41345
https://bugs.python.org/msg374466

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/openstack/rally-openstack/+/813742

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/openstack/rally-openstack/+/813742
Committed: https://opendev.org/openstack/rally-openstack/commit/0ec255aad23aea7693f9b830962370a0500767b9
Submitter: "Zuul (22348)"
Branch: master

commit 0ec255aad23aea7693f9b830962370a0500767b9
Author: Nobuto Murata <email address hidden>
Date: Wed Oct 13 14:30:29 2021 +0900

    Avoid poll_interval to be the same as HTTP Keep-Alive timeout

    When OpenStack API services are behind Apache2, HTTP Keep-Alive timeout
    is 5 seconds out of the box. Polling operations with the same interval
    as HTTP Keep-Alive timeout can cause intermittent failures like:

    Unable to establish connection to
    https://nova.example.com:8774/v2.1/servers/UUID:
    ('Connection aborted.', RemoteDisconnected('Remote end closed connection
    without response',))

    Let's avoid the same value by default as Apache2's HTTP Keep-Alive
    timeout.

    Closes-Bug: #1946912
    Change-Id: Ibda414b129a44d38ed3c3a4b5a43fd45e63ec122

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/rally-openstack 2.2.0 release.