```

In case of a S3 multi-part operation each part is en/decrypted with the scheme shown in Figure 1. However, for each part an unique secret key is derived from the OEK and the part number using a PRF. So in case of multi-part not the OEK but the output of `PRF(OEK, part_id)` is used as secret key.

#### Cryptographic Primitives

- one end of the socket is in the application pod
- and the other end is in ztunnel's pod

and setting up iptables rules to funnel traffic thru that socket "tube" to ztunnel and back.

This effectively behaves like ztunnel is an in-pod sidecar, without actually requiring the injection of ztunnel as a sidecar into the pod manifest, or mutatating the application pod in any way.

## Concepts

- All Buckets on MinIO are always in one of the following states: unversioned (the default) and all other existing deployments, versioning-enabled, or versioning-suspended.
- Versioning state applies to all of the objects in the versioning enabled bucket. The first time you enable a bucket for versioning, objects in the bucket are thereafter always versioned and given a unique version ID.

```

### Upgrading the Chart

You can use Helm to update MinIO version in a live release. Assuming your release is named as `my-release`, get the values using the command:

```bash
helm get values my-release > old_values.yaml
```

Then change the field `image.tag` in `old_values.yaml` file with MinIO image tag you want to use. Now update the chart using

```bash

delay=15.0`. By default the value is `10.0`. This means the scanner will sleep *10x* the time each operation takes.

In most setups this will keep the scanner slow enough to not impact overall system performance. Setting the `delay` key to a *lower* value will make the scanner faster and setting it to 0 will make the scanner run at full speed (not recommended in production). Setting it to a higher value will make the scanner slower, consuming less resources with the trade off of not collecting...

The CLI and controller share the same API and codebase for generating manifests from the API. You can think of the
controller as the CLI command `istioctl install` running in a loop in a pod in the cluster and using the config
from the in-cluster IstioOperator custom resource (CR).
There are two major differences:

1. The controller does not accept any dynamic user config through flags. All user interaction is through the

Replicating Deletes

Delete marker replication is allowed in [AWS V1 Configuration](https://aws.amazon.com/blogs/storage/managing-delete-marker-replication-in-amazon-s3/) but not in V2 configuration. The MinIO implementation above is based on V2 configuration, however it has been extended to allow both DeleteMarker replication and replication of versioned deletes with the `DeleteMarkerReplication` and `DeleteReplication` fields in the replication configuration above. By default, this is set to...

### Upgrade Checklist

- Test all upgrades in a lower environment (DEV, QA, UAT) before applying to production. Performing blind upgrades in production environments carries significant risk.

mapreduce.job.reduce.slowstart.completedmaps=0.99 # 99% map, then reduce
mapreduce.reduce.shuffle.input.buffer.percent=0.9 # Min % buffer in RAM
mapreduce.reduce.shuffle.merge.percent=0.9 # Minimum % merges in RAM
mapreduce.reduce.speculative=false # Disable speculation for reducing
mapreduce.task.io.sort.factor=999 # Threshold before writing to disk

researchers push the state-of-the-art in ML and developers easily build and
deploy ML-powered applications.

TensorFlow was originally developed by researchers and engineers working within
the Machine Intelligence team at Google Brain to conduct research in machine
learning and neural networks. However, the framework is versatile enough to be
used in other areas as well.