Directly after modifying a ResourceClaim in the apiserver, the locally cached
copy is outdated until the informer receives the update. If any operation looks
at the claim during that time frame, it will act based on stale
information. For example, it might try to allocate again. If that works because
of idempotency, then the following update operation fails with a conflict
error.
This is harmless, but leads to confusing log output. It can be avoided by
keeping a copy of the updated claim and using that instead of the one from the
informer cache.
These helper packages implement the parts of a dynamic resource allocation
driver that are generic and can be used by the in-tree test driver as well as
out-of-tree vendor drivers.
Functional options make it possible to treat some parameters as
optional (logger) and extend the API later on.
Dependencies need to be updated to use
github.com/container-orchestrated-devices/container-device-interface.
It's not decided yet whether we will implement Topology support
for DRA or not. Not having any toppology-related code
will help to avoid wrong impression that DRA is used as a hint
provider for the Topology Manager.
due to redesign where we changed from new endpoint to /apis. The expected error was not also changed.
Caught by e2e tests when feature enabled. Should have been caught by unit test first but it was implemented without root /apis. Unit test also fixed
This is similar to the support code for generic ephemeral inline volumes.
Differences:
- to avoid stuttering, the functions are just resourceclaim.Name and
resourceclaim.IsForPod
- resourceclaim.Name returns the right name for both cases (template
and reference), which will simplify some code
Other helper functions check some simple status conditions.
This will be used for different individual packages. A separate repo
is better than reusing something like component-helpers because dependencies
will be different.
This adds a new resource.k8s.io API group with v1alpha1 as version. It contains
four new types: resource.ResourceClaim, resource.ResourceClass, resource.ResourceClaimTemplate, and
resource.PodScheduling.
Also make some design changes exposed in testing and review.
Do not remove the ambiguous old metric
`apiserver_flowcontrol_request_concurrency_limit` because reviewers
though it is too early. This creates a problem, that metric can not
keep both of its old meanings. I chose the configured concurrency
limit.
Testing has revealed a design flaw, which concerns the initialization
of the seat demand state tracking. The current design in the KEP is
as follows.
> Adjustment is also done on configuration change … For a newly
> introduced priority level, we set HighSeatDemand, AvgSeatDemand, and
> SmoothSeatDemand to NominalCL-LendableSD/2 and StDevSeatDemand to
> zero.
But this does not work out well at server startup. As part of its
construction, the APF controller does a configuration change with zero
objects read, to initialize its request-handling state. As always,
the two mandatory priority levels are implicitly added whenever they
are not read. So this initial reconfig has one non-exempt priority
level, the mandatory one called catch-all --- and it gets its
SmoothSeatDemand initialized to the whole server concurrency limit.
From there it decays slowly, as per the regular design. So for a
fairly long time, it appears to have a high demand and competes
strongly with the other priority levels. Its Target is higher than
all the others, once they start to show up. It properly gets a low
NominalCL once other levels show up, which actually makes it compete
harder for borrowing: it has an exceptionally high Target and a rather
low NominalCL.
I have considered the following fix. The idea is that the designed
initialization is not appropriate before all the default objects are
read. So the fix is to have a mode bit in the controller. In the
initial state, those seat demand tracking variables are set to zero.
Once the config-producing controller detects that all the default
objects are pre-existing, it flips the mode bit. In the later mode,
the seat demand tracking variables are initialized as originally
designed.
However, that still gives preferential treatment to the default
PriorityLevelConfiguration objects, over any that may be added later.
So I have made a universal and simpler fix: always initialize those
seat demand tracking variables to zero. Even if a lot of load shows
up quickly, remember that adjustments are frequent (every 10 sec) and
the very next one will fully respond to that load.
Also: revise logging logic, to log at numerically lower V level when
there is a change.
Also: bug fix in float64close.
Also, separate imports in some file
Co-authored-by: Han Kang <hankang@google.com>
