Description of problem: Sometimes when setting a memory limit in a DeploymentConfig such as 1.8G, it gets converted to a millibyte value. limits: cpu: 490m memory: 1932735283200m The following docs change removed the "m" suffix as it is useless because we cannot manage memory at millibyte granularity, so do not list the suffix as allowed. The value that gets converted in the DeploymentConfig should match those listed as valid in the docs. [2] [1] https://github.com/openshift/openshift-docs/commit/1f7090ffc2f9d4ab6d362965bf4999a0b1809e2d [2] https://docs.openshift.com/container-platform/3.5/dev_guide/compute_resources.html#dev-compute-resources Version-Release number of selected component (if applicable): openshift v3.5.5.26 kubernetes v1.5.2+43a9be4 etcd 3.1.0 How reproducible: Sometimes. it may depend on the value being entered. Steps to Reproduce: 1. oc edit dc <dc> 2. Set resources: limits: memory: 1.8G 3. oc get dc <dc> -o yaml | grep memory memory: 1932735283200m Actual results: memory: 1932735283200m Expected results: If it must be converted to a non decimal value, then it should be the next logical suffix. memory: 1800M Additional info: Customer see this in 3.2 but it is reproducible in 3.5
I have sent this PR https://github.com/kubernetes/kubernetes/pull/49732 to address this issue.
Still being discussed upstream for fix
I'm lowering the severity on this since it doesn't cause a functional issue, just an aesthetic issue when reading the spec back out. The workarounds are trivial: either use BinarySI units (Gi,Mi,Ki) or don't use decimal quantities when using DecimalSI unit (G, M, K).
Customer closed their case and this likely went out with 3.7.z. Closing as current release. If this problem still persists please open a new bug.
The needinfo request[s] on this closed bug have been removed as they have been unresolved for 1000 days