*A noticeable trait of this expansion is that it chooses unique hosts such the setup provides maximum protection and availability.*

- Choosing an erasure set for the object is decided during `PutObject()`, object names are used to find the right erasure set using the following pseudo code.

```go
// hashes the key returning an integer.
func sipHashMod(key string, cardinality int, id [16]byte) int {
        if cardinality <= 0 {
                return -1

  // items is the list of VolumeAttachments
  repeated VolumeAttachment items = 2;
}

// VolumeAttachmentSource represents a volume that should be attached.
// Right now only PersistenVolumes can be attached via external attacher,
// in future we may allow also inline volumes in pods.
// Exactly one member can be set.
message VolumeAttachmentSource {

This directory contains the helm chart _sources_ which are versioned, built and pushed to following helm
repositories with each Istio release.  If you want to make changes to Istio helm charts, you're in the
right place.

If you want to _install_ Istio with Helm, instead please [follow the Helm installation docs here](https://istio.io/latest/docs/setup/install/helm/).

Charts in this folder are published to the following Helm repos:

verify-healing-inconsistent-versions: install-race ## verify resolving inconsistent versions
	@echo "Verify resolving inconsistent versions build with race"
	@(env bash $(PWD)/buildscripts/resolve-right-versions.sh)

build-debugging:
	@(env bash $(PWD)/docs/debugging/build.sh)

build: checks build-debugging ## builds minio to $(PWD)
	@echo "Building minio binary to './minio'"

	if !retentionCfg.LockEnabled {
		if legalHoldRequested || retentionRequested {
			return mode, retainDate, legalHold, ErrInvalidBucketObjectLockConfiguration
		}

		// If this not a WORM enabled bucket, we should return right here.
		return mode, retainDate, legalHold, ErrNone
	}

	opts, err := getOpts(ctx, rq, bucket, object)
	if err != nil {
		return mode, retainDate, legalHold, toAPIErrorCode(ctx, err)
	}

			if !meta.Deleted {
				if len(meta.Erasure.Distribution) != len(onlineDisks) {
					// Erasure distribution is not the same as onlineDisks
					// attempt a fix if possible, assuming other entries
					// might have the right erasure distribution.
					partsMetadata[i] = FileInfo{}
					metaErrs[i] = errFileCorrupt
					continue
				}
			}
		}
	}

	// Copy meta errors to part errors
	for i, err := range metaErrs {

		// array second dimension represents list of disks used per set.
		Sets [][]string `json:"sets"`
		// Distribution algorithm represents the hashing algorithm
		// to pick the right set index for an object.
		DistributionAlgo string `json:"distributionAlgo"`
	} `json:"xl"`
}

// formatErasureV3 struct is same as formatErasureV2 struct except that formatErasureV3.Erasure.Version is "3" indicating

				p.buf[bufIdx] = make([]byte, p.shardSize)
			}
			// For the last shard, the shardsize might be less than previous shard sizes.
			// Hence the following statement ensures that the buffer size is reset to the right size.
			p.buf[bufIdx] = p.buf[bufIdx][:p.shardSize]
			n, err := rr.ReadAt(p.buf[bufIdx], p.offset)
			if err != nil {
				switch {
				case errors.Is(err, errFileNotFound):

            this.parameterOffset += this.pad1;

            this.totalParameterCount = writeParametersWireFormat(this.txn_buf, this.bufParameterOffset);
            this.bufDataOffset = this.totalParameterCount; // data comes right after data

            int available = this.maxBufferSize - this.parameterOffset;
            this.parameterCount = Math.min(this.totalParameterCount, available);
            available -= this.parameterCount;

#### Blocking APIs

Blocking APIs are convenient because you get top-to-bottom procedural code without indirection. Network calls work like regular method calls: ask for data and it is returned. If the request fails, you get a stacktrace right where the call was made.

Blocking APIs may be inefficient because you hold a thread idle while waiting on the network. Threads are expensive because they have both a memory overhead and a context-switching overhead.