feat: Introduced storage/size based global, container, pod, and namespace-level policies #6
Conversation
|
I noticed that a lot of the code in the controller was repetitive, barring slight differences (container vs pod vs namespace). I think it would make sense for us to generate the file instead of writing it by hand, both to preserve developer sanity and to enable easier verification of the behaviour. In go-criu, we generate the bindings for CRIU's magic values here. Maybe something similar can be used here? Also, I would pick Python over Go for this, since it's just simpler. |
Parthiba-Hazra
force-pushed
the
storage-quota
branch
17 times, most recently
from
63947e4 to
3e1543a
Compare
Parthiba-Hazra
force-pushed
the
storage-quota
branch
4 times, most recently
from
5f3bc75 to
7a799f3
Compare
Parthiba-Hazra
changed the title
Parthiba-Hazra
force-pushed
the
storage-quota
branch
from
7a799f3 to
aec27f5
Compare
|
@Parthiba-Hazra Would you be able to rebase this pull request on the main branch? |
Parthiba-Hazra
force-pushed
the
storage-quota
branch
3 times, most recently
from
0f4f3aa to
8c53d87
Compare
| # containerPolicies: | ||
| # - namespace: <namespace> | ||
| # pod: <pod_name> | ||
| # container: <container_name> | ||
| # maxCheckpoints: 5 | ||
| # maxCheckpointSize: 10 | ||
| # maxTotalSize: 100 |
There was a problem hiding this comment.
It might be better to use format similar to other resource limits in Kubernetes:
https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/#meaning-of-memory
Limits and requests for memory are measured in bytes. You can express memory as a plain integer or as a fixed-point number using one of these quantity suffixes: E, P, T, G, M, k. You can also use the power-of-two equivalents: Ei, Pi, Ti, Gi, Mi, Ki.
- Introduced global, container, pod, and namespace-level policies for checkpoint retention, based on storage/size limits. - Updated CRD definitions to store the storage/size based policies. - Updated the sample configuration of CheckpointRestoreOperator with storage/checkpoint-size based policies Signed-off-by: Parthiba-Hazra <[email protected]>
- Enhance generate_checkpoint_tar.sh to optionally generate tar files larger than 5MB - Update GitHub Actions workflow to test storage quota garbage collection policies Signed-off-by: Parthiba-Hazra <[email protected]>
Signed-off-by: Parthiba-Hazra <[email protected]>
…ntions - Updated size policies from int-based values to resource.Quantity for consistency with Kubernetes resource limits. - Enabled the use of units such as Ki, Mi, Gi for max checkpoint size and total size policies. Signed-off-by: Parthiba-Hazra <[email protected]>
Parthiba-Hazra
force-pushed
the
storage-quota
branch
from
8c53d87 to
c322bf4
Compare
| maxCheckpointSize resource.Quantity = resource.MustParse("1Ei") | ||
| maxTotalSizePerPod resource.Quantity = resource.MustParse("1Ei") | ||
| maxTotalSizePerContainer resource.Quantity = resource.MustParse("1Ei") | ||
| maxTotalSizePerNamespace resource.Quantity = resource.MustParse("1Ei") |
There was a problem hiding this comment.
I'm not sure if 1 exabyte is a very sane default value. We could potentially restrict this to something a little more reasonable, like 100 gigabytes or so? cc @rst0git
Introduced storage/size based global, container, pod, and namespace-level policies for checkpoint retention.
maxCheckpointSize,maxTotalSizePerContainer) are defined in Megabyte (MB)