Bug 1474744 - [paid][online-stg] The user can use non-integer GiB to request more storages than it is limited in the storage quota.
[paid][online-stg] The user can use non-integer GiB to request more storages ...
Product: OpenShift Container Platform
Classification: Red Hat
Component: RFE (Show other bugs)
Unspecified Unspecified
high Severity medium
: ---
: ---
Assigned To: Paul Weil
Xiaoli Tian
: OnlinePro
Depends On:
Blocks: PPCTracker
  Show dependency treegraph
Reported: 2017-07-25 06:17 EDT by yasun
Modified: 2017-09-14 16:26 EDT (History)
11 users (show)

See Also:
Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of:
Last Closed:
Type: Bug
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)

  None (edit)
Comment 1 Abhishek Gupta 2017-08-22 12:33:31 EDT
Pushing this out as the user can at most shoot over by 1Gi.
Comment 3 Abhishek Gupta 2017-09-12 17:01:47 EDT
Some options I can think of on top of my head:

1. we could prevent users from specifying non-integer values
2. we could switch to using status values for quota 
3. we could round up to the next integer to determine "request" for quota purposes
Comment 4 Seth Jennings 2017-09-12 17:05:29 EDT
Did intend to reassign.  Changing back.
Comment 5 Jan Safranek 2017-09-13 04:05:46 EDT
From storage point of view, different backends have different rules. AWS, GCE and Cinder can allocate only whole GB (and Kubernetes rounds non-integer GBs up), while e.g. Gluster can allocate in MB. We could force all of them to round to GB, at least in quota calculations.

It would be nice if we could account also pvc.status.capacity which always holds the size of bound PV. Until a PVC is bound its spec.capacity would be used for quota and after binding its status.capacity would be used instead, as it reflects real size of allocated storage to the user.
Comment 6 Eric Paris 2017-09-14 13:30:16 EDT
Assigning to brad. This is a storage 'problem' not online specific. Jan has an idea. This might be an RFE. I'll let Brad and storage team decide how to run with this.

Note You need to log in before you can comment on or make changes to this bug.