log: add warn log when resource request greater than limit

[ZYecho]

Signed-off-by: zhangyue zy675793960@yeah.net
add warn log when memory resource request greater than limit
fix #418

[sdudoladov]

Please also add this warning to the code that configures sidecars and init containers as they are also affected. Note that a Scalyr sidecar is configured separately.

[ZYecho]

Please also add this warning to the code that configures sidecars and init containers as they are also affected. Note that a Scalyr sidecar is configured separately.

they already use RequestIsSmallerThanLimit to judge and log a warn, what else do you want?

[sdudoladov]

Please also add this warning to the code that configures sidecars and init containers as they are also affected. Note that a Scalyr sidecar is configured separately.

they already use RequestIsSmallerThanLimit to judge and log a warn, what else do you want?

logging a warning when a resource request as seen by the operator is greater than the limit for all the containers that an operator can potentially create. This includes up to 4 container types:

1. Postgres container - already covered by this PR
2. Scalyr sidecar
3. init containers
4. other sidecars

the whole idea is to make visible in the logs a situation where a container cannot start because the memory request > limit . Right now finding this is the case involves running kubectl describe on various k8s resources. Logging such misconfiguration shall improve the operational experience for the operator users.

[sdudoladov]

That won't work because that will reject the legitimate case of request being equal to limit.

In RequestIsSmallerThanLimit, the line

 return request.Cmp(limit) == -1, nil

is responsible for that . request.Cmp(limit) will return zero if quantities are equal. Please check Kubernetes source code / docs for that method.