Fix volume size allocation in gcd_pd #56600

edisonxiang · 2017-11-30T01:33:41Z

What this PR does / why we need it:
GCE PDs are allocated in chunks of GBs not GiB but CreateVolume function incorrectly creates volume in chunks of GiB.
1 GiB = 1024 * 1024 * 1024 Bytes
1 GB = 1000 * 1000 * 1000 Bytes

Which issue(s) this PR fixes (optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close the issue(s) when PR gets merged):
Fixes #56081

Special notes for your reviewer:

Release note:

Fixed dynamic provisioning of GCE PDs to round to the next GB instead of GiB

msau42 · 2017-11-30T01:38:34Z

/ok-to-test

edisonxiang · 2017-11-30T01:41:34Z

/cc @msau42 @gnufied

gnufied · 2017-11-30T01:45:08Z

pkg/volume/gce_pd/gce_util.go

@@ -194,7 +196,7 @@ func (gceutil *GCEDiskUtil) CreateVolume(c *gcePersistentDiskProvisioner) (strin
 		glog.Errorf("error getting labels for volume %q: %v", name, err)
 	}

-	return name, int(requestGB), labels, fstype, nil
+	return name, int(requestGiB), labels, fstype, nil


Lets make sure that value returned by this function is read correctly by Provision function here - https://github.com/kubernetes/kubernetes/blob/master/pkg/volume/gce_pd/gce_pd.go#L426 . Currently as it exists it will try to save a GB value back into GiB and that will be wrong.

Ah, I missed that you are creating both GB and GiB values out of requested size and returning GiB value back to caller. That works too I think.

@gnufied Thanks your review.

Would be worth renaming the variables GB here:
https://github.com/kubernetes/kubernetes/blob/master/pkg/volume/gce_pd/gce_pd.go#L405
and here:
https://github.com/kubernetes/kubernetes/blob/master/pkg/volume/gce_pd/gce_pd.go#L426

@davidz627 Done, thanks.

I'm confused, why do we need to return GiB if we're allocating in GB? Won't the size recorded in the PV be wrong then?

Hi @msau42, Thanks your review. Does your suggestion is to remove this line and return requestGB directly, and replace "%dGi" with "%dG" in this line?

I think either thing works. Return the GB value and then get resource.Quantity out of it by doing %dG. That is what I originally suggested above. Returning GiB and leaving %dGi works too, but I do think that may be returning GB and then parsing value back using %dG was better.

I don't think returning GiB is accurate. requestGiB != requestGB in bytes. The PV is going to have capacity larger than what was actually allocated.

So if there is no reason to return values in Gi, then I would recommend just returning GB.

@msau42 @gnufied Done. Now it is return GB Directly. Thanks.

edisonxiang · 2017-12-01T01:49:21Z

Hey @davidz627 @msau42 @gnufied PTAL, Thank you~

davidz627 · 2017-12-01T22:02:42Z

/lgtm

gnufied · 2017-12-01T22:07:46Z

/approve

msau42 · 2017-12-04T17:51:01Z

/lgtm cancel

davidz627 · 2017-12-05T01:48:12Z

pkg/volume/util.go

@@ -38,6 +38,13 @@ import (
 	"k8s.io/apimachinery/pkg/util/sets"
 )

+const (
+	// GiB is used by PV/PVC. 1 GiB = 1024 * 1024 * 1024 Bytes
+	VolumeSizeGiB = 1024 * 1024 * 1024


are we using this variable anywhere?

Yeah. A lot of places. Such as this line. If this bug fix is accepted, I am willing to submit another pr to fix this issue with @gnufied.

msau42

/lgtm

Thanks for running the e2e tests too!

fejta-bot · 2017-12-09T12:24:47Z

/retest
This bot automatically retries jobs that failed/flaked on approved PRs (send feedback to @fejta).

Review the full test history for this PR.

fejta-bot · 2017-12-09T15:33:47Z

/retest
This bot automatically retries jobs that failed/flaked on approved PRs (send feedback to @fejta).

Review the full test history for this PR.

msau42 · 2017-12-11T19:00:05Z

Does this need a release note, since we will now be provisioning GCE disks that are smaller than before?

edisonxiang · 2017-12-17T11:50:24Z

@msau42 Done, PTAL, Thanks.

gnufied · 2017-12-18T15:38:00Z

pkg/volume/gce_pd/gce_util.go

-	// GCE works with gigabytes, convert to GiB with rounding up
-	requestGB := volume.RoundUpSize(requestBytes, 1024*1024*1024)
+	// GCE PDs are allocated in chunks of GBs (not GiBs)
+	requestGB := volume.RoundUpSize(requestBytes, volume.GB)



This is really tiny nit - but can we simply use volume.RoundUpToGB and remove that reuestBytes variable?

@gnufied, done, Thanks.

edisonxiang · 2017-12-19T03:09:18Z

/test pull-kubernetes-e2e-gce
/test pull-kubernetes-verify

gnufied · 2017-12-19T19:46:59Z

/lgtm

k8s-ci-robot · 2017-12-19T19:47:22Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: davidz627, edisonxiang, gnufied, msau42

Associated issue: #56081

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these OWNERS Files:

~~pkg/volume/gce_pd/OWNERS~~ [gnufied]
~~test/e2e/storage/OWNERS~~ [gnufied,msau42]

You can indicate your approval by writing /approve in a comment
You can cancel your approval by writing /approve cancel in a comment

k8s-github-robot · 2017-12-19T20:37:45Z

/test all [submit-queue is verifying that this PR is safe to merge]

k8s-github-robot · 2017-12-19T22:08:37Z

Automatic merge from submit-queue (batch tested with PRs 56600, 56814). If you want to cherry-pick this change to another branch, please follow the instructions here.

… GiB instead of GBs Fixing comments and importing constant from util Importing constant from util Fixing comment in volume_provisioning.go

…ing_to_GB Automatic merge from submit-queue (batch tested with PRs 66172, 66254). If you want to cherry-pick this change to another branch, please follow the instructions <a href="https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md">here</a>. Reverting commit #56600 as GCE PD is allocated in chunks of GiB inste… **What this PR does / why we need it:** This PR reverts the changes made in commit #56600 which considered GCE PDs are allocated in chunks of GBs. The following set of operations demonstrate the allocation is in GiBs. Manually create a PD in GB, and manually attach it to a node: ``` $ gcloud compute disks create msau-test --zone=us-central1-b --size=1GB ``` Run lsblk on it, and it shows the bytes of the disk are 1GiB: ``` $ lsblk -b sdc 8:32 0 1073741824 0 disk ``` **Which issue(s) this PR fixes**: [65285](#65285) **Special notes for your reviewer**: ```release-note none ```

… GiB instead of GBs Fixing comments and importing constant from util Importing constant from util Fixing comment in volume_provisioning.go

…6172-#66324-upstream-release-1.10 Automatic merge from submit-queue. Automated cherry pick of #66172: Reverting commit #56600 as GCE PD is allocated in chunks of #66324: Fixing E2E tests for disk resizing Cherry pick of #66172 #66324 on release-1.10. #66172: Reverting commit #56600 as GCE PD is allocated in chunks of #66324: Fixing E2E tests for disk resizing ```release-note none ```

…ated in chunks of GiB inste... Origin-commit: 5f297b287dee3d96b4a0541c348832a330772713

…6172-#66324-upstream-release-1.11 Automatic merge from submit-queue. Automated cherry pick of #66172: Reverting commit #56600 as GCE PD is allocated in chunks of #66324: Fixing E2E tests for disk resizing Cherry pick of #66172 #66324 on release-1.11. #66172: Reverting commit #56600 as GCE PD is allocated in chunks of #66324: Fixing E2E tests for disk resizing ```release-note none ```

… GiB instead of GBs Fixing comments and importing constant from util Importing constant from util Fixing comment in volume_provisioning.go

…ated in chunks of GiB inste... Origin-commit: 5f297b287dee3d96b4a0541c348832a330772713

k8s-github-robot assigned verult and davidz627 Nov 30, 2017

edisonxiang mentioned this pull request Nov 30, 2017

GCE CreateVolume allocates in chunks of GiB incorrectly #56081

Closed

k8s-ci-robot removed the needs-ok-to-test Indicates a PR that requires an org member to verify it is safe to test. label Nov 30, 2017

k8s-ci-robot requested review from gnufied and msau42 November 30, 2017 01:41

gnufied reviewed Nov 30, 2017

View reviewed changes

k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Dec 1, 2017

k8s-github-robot added the approved Indicates a PR has been approved by an approver from all required OWNERS files. label Dec 1, 2017

k8s-ci-robot assigned msau42 Dec 4, 2017

k8s-ci-robot removed the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Dec 4, 2017

davidz627 reviewed Dec 5, 2017

View reviewed changes

k8s-ci-robot added size/M Denotes a PR that changes 30-99 lines, ignoring generated files. and removed size/S Denotes a PR that changes 10-29 lines, ignoring generated files. labels Dec 5, 2017

msau42 reviewed Dec 5, 2017

View reviewed changes

k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Dec 5, 2017

k8s-ci-robot added the release-note Denotes a PR that will be considered when it comes time to generate release notes. label Dec 14, 2017

edisonxiang force-pushed the fixvolumesize branch from 0b4ff6e to d80dbe7 Compare December 17, 2017 11:48

gnufied reviewed Dec 18, 2017

View reviewed changes

using RoundUpToGB function directly

d9dc642

k8s-ci-robot assigned gnufied Dec 19, 2017

k8s-ci-robot added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Dec 19, 2017

k8s-github-robot merged commit 7ede2a0 into kubernetes:master Dec 19, 2017

edisonxiang deleted the fixvolumesize branch January 30, 2018 08:21

jsafrane mentioned this pull request Mar 5, 2018

[test failed] [1.10 upgrade] Dynamic Provisioning DynamicProvisioner #60766

Closed

msau42 mentioned this pull request Jun 20, 2018

GCE PD actually rounds to GiB #65285

Closed

krunaljain mentioned this pull request Jul 17, 2018

Reverting commit #56600 as GCE PD is allocated in chunks of GiB inste… #66172

Merged

krunaljain mentioned this pull request Jul 19, 2018

Automated cherry pick of #66172: Reverting commit #56600 as GCE PD is allocated in chunks of #66324: Fixing E2E tests for disk resizing #66355

Merged

krunaljain mentioned this pull request Jul 19, 2018

Automated cherry pick of #66172: Reverting commit #56600 as GCE PD is allocated in chunks of #66324: Fixing E2E tests for disk resizing #66356

Merged

openshift-publish-robot pushed a commit to openshift/kubernetes that referenced this pull request Jul 29, 2018

UPSTREAM: 66172: Reverting commit kubernetes#56600 as GCE PD is alloc…

360ba42

…ated in chunks of GiB inste... Origin-commit: 5f297b287dee3d96b4a0541c348832a330772713

openshift-publish-robot pushed a commit to openshift/kubernetes that referenced this pull request Jan 29, 2019

UPSTREAM: 66172: Reverting commit kubernetes#56600 as GCE PD is alloc…

67f0994

…ated in chunks of GiB inste... Origin-commit: 5f297b287dee3d96b4a0541c348832a330772713

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix volume size allocation in gcd_pd #56600

Fix volume size allocation in gcd_pd #56600

edisonxiang commented Nov 30, 2017 •

edited

Loading

msau42 commented Nov 30, 2017

edisonxiang commented Nov 30, 2017

gnufied Nov 30, 2017 •

edited

Loading

gnufied Nov 30, 2017

edisonxiang Nov 30, 2017

davidz627 Nov 30, 2017

edisonxiang Dec 1, 2017

msau42 Dec 1, 2017

edisonxiang Dec 4, 2017

gnufied Dec 4, 2017

msau42 Dec 4, 2017

edisonxiang Dec 5, 2017

edisonxiang commented Dec 1, 2017

davidz627 commented Dec 1, 2017

gnufied commented Dec 1, 2017

msau42 commented Dec 4, 2017

davidz627 Dec 5, 2017

edisonxiang Dec 5, 2017

msau42 left a comment

fejta-bot commented Dec 9, 2017

fejta-bot commented Dec 9, 2017

msau42 commented Dec 11, 2017

edisonxiang commented Dec 17, 2017

gnufied Dec 18, 2017

edisonxiang Dec 19, 2017

edisonxiang commented Dec 19, 2017

gnufied commented Dec 19, 2017

k8s-ci-robot commented Dec 19, 2017

k8s-github-robot commented Dec 19, 2017

k8s-github-robot commented Dec 19, 2017

Fix volume size allocation in gcd_pd #56600

Fix volume size allocation in gcd_pd #56600

Conversation

edisonxiang commented Nov 30, 2017 • edited Loading

msau42 commented Nov 30, 2017

edisonxiang commented Nov 30, 2017

gnufied Nov 30, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

edisonxiang commented Dec 1, 2017

davidz627 commented Dec 1, 2017

gnufied commented Dec 1, 2017

msau42 commented Dec 4, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

msau42 left a comment

Choose a reason for hiding this comment

fejta-bot commented Dec 9, 2017

fejta-bot commented Dec 9, 2017

msau42 commented Dec 11, 2017

edisonxiang commented Dec 17, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

edisonxiang commented Dec 19, 2017

gnufied commented Dec 19, 2017

k8s-ci-robot commented Dec 19, 2017

k8s-github-robot commented Dec 19, 2017

k8s-github-robot commented Dec 19, 2017

edisonxiang commented Nov 30, 2017 •

edited

Loading

gnufied Nov 30, 2017 •

edited

Loading