Closed ghost closed 3 years ago
In GitLab by @infinitewarp on Sep 10, 2020, 15:51
marked this issue as related to cloudigrade#728
In GitLab by @infinitewarp on Sep 10, 2020, 15:51
I'm marking this as related to https://gitlab.com/cloudigrade/cloudigrade/-/issues/728 because I think that may be the root cause of this problem.
In GitLab by @infinitewarp on Sep 17, 2020, 10:49
removed the relation with cloudigrade#728
In GitLab by @infinitewarp on Sep 17, 2020, 10:49
marked this issue as related to cloudigrade#728
In GitLab by @mpierce on Oct 6, 2020, 14:45
There are additional images that return conflicting results from inspections. Still in DEV07 account. All details about recreating behavior are the same as above.
(The details below are formatted differently in each run because I didn't realize I needed to collect it until after-the-fact and that's what I had available. I'll gather more details in another run if needed.)
ami-01eb13851a0ed6486
and
ami-0d1b0d4e138bf4cdf
More details about each image is below:
'ami-01eb13851a0ed6486': image(
ami='ami-01eb13851a0ed6486',
arch='x86_64',
role='Workstation',
type='t2.micro',
sla='Premium',
usage='Dev/Test',
serv_type='L3',
name='swatch-conc-test-1',
RHEL_expected='False'
),
test_run_2:
Arch: x86_64
AMI: ami-01eb13851a0ed6486
RHEL: True
rhel_detected: True
test_run_1:
'ami-0d1b0d4e138bf4cdf': image(
ami='ami-0d1b0d4e138bf4cdf',
arch='arm64',
role='Console',
type='a1.medium',
sla='self-support',
usage='prod',
serv_type='L3',
name='swatch-concurrent-test-4',
RHEL_expected='True'
),
test_run_2:
Arch: arm64
AMI: ami-0d1b0d4e138bf4cdf
RHEL: False
rhel_detected: False```
In GitLab by @infinitewarp on Oct 8, 2020, 09:52
assigned to @infinitewarp
In GitLab by @infinitewarp on Oct 12, 2020, 10:55
mentioned in merge request cloudigrade!806
In GitLab by @infinitewarp on Oct 12, 2020, 10:58
mentioned in merge request shiftigrade!136
In GitLab by @infinitewarp on Oct 12, 2020, 14:04
With the aforementioned cloudigrade!806, I introduced changes to cloudigrade's inspection process so that it should verify that all expected volumes are actually attached to the cluster instance before running the houndigrade task, and it should back off and retry until volumes are ready.
In GitLab by @infinitewarp on Oct 13, 2020, 09:49
assigned to @pakamble
In GitLab by @Maribeth on Sep 10, 2020, 15:28
Summary
When Houndigrade's inspection process runs against images, it occasionally returns different results. Like sometimes it reports back
RHEL=True
and sometimes it reports backRHEL=False
for instances of the same image.Steps to Reproduce
In the
qa
env, usingdev07
AWS accountTest automation fired up all these instances, created the Source and Cloud Meter accounts and waited for inspection. Once the inspection process finished, I
httpie
dcloudigrade/v2/images/#####/
to see the results of the inspection.Images used in my current test:
Expected Result
RHEL=True
vs.RHEL=False
Actual Result
Additional context
Responses from inspection process for image
ami-0169e18bfd191f4e6
(RHEL=True
)Same image,
RHEL=False
Inspection results returned
RHEL=False
for different imageami-07e90d06363f8ebd9
and
RHEL=True
from a different inspection of the same image: