Skip to content

Commit

Permalink
Add CHANGELOG.md
Browse files Browse the repository at this point in the history 
This commit adds a CHANGELOG.md file. Rather than automatically constructing
release notes from the git commit log using semantic commits, we've opted to
have dev's add notes about user facing changes to this log as they commit them.

This commit also slightly expands CONTRIBUTING.md and removes the, now very,
outdated notes.

Closes #1406
  • Loading branch information
@chrisseto @RafalKorepta
chrisseto authored and RafalKorepta committed Jul 10, 2024
1 parent cadc029 commit 43cfb87
Show file tree
Hide file tree
Showing 2 changed files with 93 additions and 93 deletions.
85 changes: 85 additions & 0 deletions CHANGELOG.md
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,85 @@
# Change Log

## Redpanda Chart

### [Unreleased](https://github.com/redpanda-data/helm-charts/releases/tag/redpanda-FILLMEIN) - YYYY-MM-DD

#### Added

#### Changed
* `image.repository` longer needs to be the default value of
`"docker.redpanda.com/redpandadata/redpanda"` to respect version checks of
`image.tag`
([#1334](https://github.com/redpanda-data/helm-charts/issues/1334)).
* `post_upgrade_job.extraEnv` and `post_upgrade_job.extraEnvFrom` no longer accept string inputs.

Previously, they accepted either strings or structured fields. As the types
of this chart are reflected in the operator's CRD, we are bound by the
constraints of Kubernetes' CRDs, which do not support fields with multiple
types. We also noticed that the [CRD requires these fields to be structured
types](https://github.com/redpanda-data/redpanda-operator/blob/9fa7a7848a22ece215be36dd17f0e4c2ba0002f7/src/go/k8s/api/redpanda/v1alpha2/redpanda_clusterspec_types.go#L597-L600)
rather than strings. Too minimize the divergences between the two, we've
opted to drop support for string inputs here but preserve them elsewhere.

Updating these fields, if they are strings, is typically a case of needing
to remove `|-`'s from one's values file.

Before:
```yaml
post_upgrade_job:
extraEnv: |-
- name: SPECIAL_LEVEL_KEY
valueFrom:
configMapKeyRef:
name: special-config
key: special.how
```
After:
```yaml
post_upgrade_job:
extraEnv:
- name: SPECIAL_LEVEL_KEY
valueFrom:
configMapKeyRef:
name: special-config
key: special.how
```
If you were using a templated value and would like to see it added back,
please [file us an
issue](https://github.com/redpanda-data/helm-charts/issues/new/choose) and
tell us about your use case!
#### Fixed
* Numeric node/broker configurations are now properly transcoded as numerics.
#### Removed
## Redpanda Operator Chart
### [Unreleased](https://github.com/redpanda-data/helm-charts/releases/tag/operator-FILLMEIN) - YYYY-MM-DD
#### Added
#### Changed
#### Fixed
#### Removed
## Connectors Chart
### [Unreleased](https://github.com/redpanda-data/helm-charts/releases/tag/connectors-FILLMEIN) - YYYY-MM-DD
#### Added
#### Changed
#### Fixed
#### Removed
## Console Chart
### [Unreleased](https://github.com/redpanda-data/helm-charts/releases/tag/console-FILLMEIN) - YYYY-MM-DD
#### Added
#### Changed
#### Fixed
#### Removed
## Kminion Chart
### [Unreleased](https://github.com/redpanda-data/helm-charts/releases/tag/console-FILLMEIN) - YYYY-MM-DD
#### Added
#### Changed
#### Fixed
#### Removed
101 changes: 8 additions & 93 deletions CONTRIBUTING.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -27,98 +27,13 @@ nix develop -c fish # Enter a development shell using fish/zsh
nix develop -c zsh # Enter a development shell using fish/zsh
```

## Contributing
## Making Changes

One way to debug during helm development is to diff the kubernetes configuration files that helm generates:

### Setup

Create and initialize git in a new directory (make sure to be outside an existing source-controlled directory):

```sh
mkdir helm-output
cd helm-output
git init
```

#### Create scripts

Create a new file `split-output.sh` in the `helm-output` directory:

```
#!/bin/bash
# split-output.sh - split helm output into individual files
if (($# != 2)); then
echo "Requires the helm directory and namespace as arguments"
echo "./split-output.sh <helm directory> <namespace>"
echo "ex: ./split-output.sh helm-charts/redpanda helm-test"
exit 1
fi
helm -n $2 install redpanda $1 --dry-run > input.yaml
csplit -s --suppress-matched -z input.yaml /---/ '{*}'
# remove the first file (it is helm metadata rather than a k8s object)
rm xx00
# loop through each file and rename according to to k8s name and source
# ex. <name>-<source-file>, or redpanda-statefulset.yaml
for file in xx*; do
NEWNAME=`grep -Po "(?<=^\ \ name:\ ).*" $file | sed 's/"//g' | xargs`-`head -1 "$file" | cut -d '/' -f 3 | sed 's/\.yaml//'`.yaml
mv "$file" $NEWNAME
done
```

Create another new file `get-redpanda-config.sh` in the same directory:

```
#!/bin/bash
# get-redpanda-config.sh - retrieves Redpanda config from a running node
# This node's configuration is printed out at startup, and a file responsible for this is application.cc
# We need to find the relevant message in the log to then filter all subsequent messages by
if (($# != 1)); then
echo "Requires namespace as the first argument"
echo "./get-redpanda-config.sh <namespace>"
echo "ex: ./get-redpanda-config.sh helm-test"
exit 1
fi
COUNT=$(kubectl -n $1 get sts | sed '2q;d' | awk '{print $2}' | cut -c1-1)
for (( i=0; i<$COUNT; i++)); do
RELEVANTLINE=$(kubectl logs -n $1 redpanda-$i | grep application.cc | sed '7q;d' | awk '{print $8}')
kubectl logs -n $1 redpanda-$i | grep $RELEVANTLINE | awk -F' - ' '{ $1=""; $2=""; print}' > ~/projects/redpanda/helm-output/redpanda-$i-config.txt
done
```

Make both of these files executable:

```sh
chmod +x split-output.sh && chmod +x get-redpanda-config.sh
```

#### Run scripts

Collect the initial logs and yaml files from a running Redpanda cluster. The namespace `helm-test` is used below... replace with whatever namespace you used for your redpanda install. This will be the baseline of your output files, so you will be able to compare any differences based on future changes to the helm chart:

```sh
./get-redpanda-config.sh helm-test
./split-output.sh ../helm-charts/redpanda helm-test
```

You will have many yaml files along with a `redpanda-0-config.txt` file and your two scripts. Commit all these files to the repo:

```sh
git add . && git commit -m 'init commit'
```

### Iterate

Now you are ready to make changes to the helm chart and eventually restart your cluster. Once this is done, re-run the `get-redpanda-config.sh` and `split-output.sh` scripts to extract the same files again, and use git to diff the results.

#### Restoring output directory

One you are done comparing the differences, you will likely want to get back to the initial state for these log files so you can compare against subsequent runs. Do this with the following command:


```sh
git restore -- . && git clean -df
```

But be careful! The above command will throw away all changes in whatever git repo you run it against, and you will be back to the most recent commit.
User facing changes should be documented in [the CHANGELOG][./CHANGELOG.md]
under the chart appropriate subheading. Changes are grouped into "Added,
Changed, Fixed, and Removed". Breaking changes[^1] and deprecations should be
prefixed with the "BREAKING" and "DEPRECATED", respectively.

[^1]: A change is considered breaking if the `values.yaml`, sans deprecated fields,
from the previous release cannot be used in a `helm upgrade` command without
modifications.

0 comments on commit 43cfb87

Please sign in to comment.