-
Notifications
You must be signed in to change notification settings - Fork 238
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
disk: DescribeDisks auto batching #1009
base: master
Are you sure you want to change the base?
Conversation
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: huww98 The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Hi @huww98. Thanks for your PR. I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
TODO: the ecsClient in waiter cannot update its credential. |
0453ea3
to
d932b08
Compare
I have observed that after AttachDisk returns, the status remains Available for a short period, so I'm not relying on disk.Status in this version. |
d1d8cbc
to
fc295d9
Compare
/ok-to-test |
In this version, I issue exactly one DescribeDisk request every 2s. So, if the number of pending disks goes just above 100, the QPS will not be doubled suddenly. If 300 disks are pending, then the status of each disk will be updated every 6s, and the QPS of DescribeDisk should remains constant. |
/auto-cc |
3f695ee
to
ad5a9d5
Compare
ad5a9d5
to
f1949b4
Compare
f1949b4
to
45ee17a
Compare
A central place for all disks pending status changes (attaching/detaching). Waiter will poll disk status in batch and wake up the caller if the status has changed. The motivation is reducing OpenAPI call count, to avoid triggering throttling, thus improve system throughput. Also include a simple fallback waiter to support multi-tanent use-case. Shared disk support is best effort, since we cannot create new shared disks for test now. Note: We have observed that after AttachDisk returns, the status remains Available for a short period, so don't rely on disk.Status in pred.
introduce two functions: waitForDisk{Attached,Detached} replace original two waitFor*DiskInStatus(), and two similar wait loop in detach*Disk()
should be reverted once we have a ecsClient that can refresh its credentials
45ee17a
to
21f3a26
Compare
What type of PR is this?
/kind feature
/kind cleanup
What this PR does / why we need it:
disk: introduce DiskStatusWaiter
A central place for all disks pending status changes (attaching/detaching).
Waiter will poll disk status in batch and wake up the caller if the status has changed.
The motivation is reducing OpenAPI call count, to avoid triggering throttling, thus improve system throughput.
Also include a simple fallback waiter to support multi-tanent use-case.
Shared disk support is best effort, since we cannot create new shared disks for test now.
Which issue(s) this PR fixes:
Fixes #
Special notes for your reviewer:
Does this PR introduce a user-facing change?
Additional documentation e.g., KEPs (Kubernetes Enhancement Proposals), usage docs, etc.: