Skip to content

shorter readme - too many words #1

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.

Sign up for GitHub

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

Open
jkremser wants to merge 1 commit into
base: main
Choose a base branch
from
Open

shorter readme - too many words #1

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 1 addition & 35 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,10 +3,7 @@
KEDR is a fast, standalone Kubernetes resource recommender written entirely in Go. It reads workload configuration from Kubernetes, queries a Prometheus-compatible metrics service, and produces CPU and memory request/limit recommendations.

KEDR uses [`github.com/kedify/recommender/analysis`](https://github.com/kedify/recommender)
for all recommendation calculations through `analysis.Analyze`. Its CLI and report
layout originated as a clean-room port of [Robusta KRR](https://github.com/robusta-dev/krr),
but its recommendation semantics are no longer KRR-compatible. It has no Python
runtime or source dependency.
for all recommendation calculations through `analysis.Analyze`.

## Demo

Expand Down Expand Up @@ -47,13 +44,6 @@ go build -o bin/kedr ./cmd/kedr
./bin/kedr simple
```

Many existing KRR flags still work, including underscore aliases:

```sh
alias krr=kedr
krr simple --namespace default --formatter json --logtostderr
```

Use `kedr simple --help` or `kedr simple_limit --help` for the complete option reference.

Table output includes a saved-run reference and an `explain` hint. Add `--explain`
Expand Down Expand Up @@ -147,13 +137,6 @@ and bounds. `simple` retains CPU limits; `explain` does not introduce limit remo
(default 15%), and a memory limit/request ratio of 1, subject to retained settings
and the analyzer's safety guards.

KEDR maps its existing CPU/memory minimums (10 millicores / 100 MiB), request
percentile, memory buffer, OOM increase, and sample-count flags into the shared
policy. CPU headroom is 1. Other bounds and material-change thresholds come from
the module. No extra KRR percentile interpolation, sizing, rounding, or floors run
after `Analyze`; exact recommendations are retained, with CPU converted from the
module's millicores to report cores. Display formatting does not change stored values.

Known unset requests and limits can receive initial values below the analyzer's
minimum-change thresholds. These thresholds apply only to existing numeric
settings. Evidence and safety guards still apply; `simple` continues to leave CPU
Expand Down Expand Up @@ -312,19 +295,6 @@ documented rollback limitations.

## Reports

The default terminal table uses the Charm Bubble Tea, Bubbles, and Lip Gloss stack. Non-terminal output never contains animation or ANSI escapes.

CPU Diff and Memory Diff use separate color scales over the visible rows. Savings
range from muted grey-green to bright green; increases range from muted warm grey
through orange to bright red. The largest change of each sign is bold. Scales use
the displayed total request change across current pods, excluding unset/unknown allocations,
unchanged displayed values, and zero totals. Remaining numeric diffs outside the
scale are grey. Equal changes share
a shade; a single change of one sign uses the brightest shade.
Colors adapt to the terminal's supported palette. `--no-color` or any nonempty
`NO_COLOR` value disables all terminal colors, including headers and the progress
indicator. Redirected output and saved table files contain no ANSI styling.

Table resource quantities are rounded to whole numbers. CPU always uses the
nearest millicore (`m`), including values above one core (`542.55m` becomes `543m`);
memory uses binary units.
Expand Down Expand Up @@ -353,10 +323,6 @@ diagnostic reasons and notices. Insufficient history includes observed and requi
hours. Credentials remain
masked in serialized configuration.

## Deliberately excluded

KEDR does not contain Robusta SaaS publishing, a web UI, HolmesGPT, Slack delivery, Azure Blob/Teams delivery, the KRR enforcer, Helm charts, or runtime-loaded custom Python strategies and formatters. Flags belonging to these removed features are rejected as unknown.

## Development

The supported toolchain is Go 1.27 or newer.
Expand Down
Loading