}

// newHTTPListener - creates new httpListener object which is interface compatible to net.Listener.
// httpListener is capable to
// * listen to multiple addresses
// * controls incoming connections only doing HTTP protocol
func newHTTPListener(ctx context.Context, serverAddrs []string, opts TCPOptions) (listener *httpListener, listenErrs []error) {
	listeners := make([]net.Listener, 0, len(serverAddrs))

- Each pool expects each host in this pool has the same number of drives specified as any other host.
- Mixing `local-path` and `distributed-path` is not allowed, doing so would cause MinIO to refuse starting the server.
- Ellipses notation (e.g. `{1...10}`) or bracket notations are fully allowed (e.g. `{a,c,f}`) to have multiple entries in one line.

	// simplicity it avoids the conversion entirely for relative
	// paths or paths containing .. elements. For now,
	// \\server\share paths are not converted to
	// \\?\UNC\server\share paths because the rules for doing so
	// are less well-specified.
	if len(path) >= 2 && path[:2] == `\\` {
		// Don't canonicalize UNC paths.
		return path
	}
	if !filepath.IsAbs(path) {
		// Relative path
		return path
	}

eam/operations/installation.html#procedure), where there are multiple pods with different domain names expected to run, you will either need wildcard certificates valid for all the domains or have specific certificates for each domain. If you are going to use specific certificates, make sure to create Kubernetes secrets accordingly.

For testing purposes, here is [how to create self-signed certificates](https://github.com/minio/minio/tree/master/docs/tls#3-generate-self-signed-certificates)....

// errDiskNotFound - cannot find the underlying configured disk anymore.
var errDiskNotFound = StorageErr("drive not found")

// errDiskOngoingReq - indicates if the disk has an on-going request in progress.
var errDiskOngoingReq = StorageErr("drive still did not complete the request")

// errDriveIsRoot - cannot use the disk since its a root disk.

- Erasure coding used by MinIO is [Reed-Solomon](https://github.com/klauspost/reedsolomon) erasure coding scheme, which has a total shard maximum of 256 i.e 128 data and 128 parity. MinIO design goes beyond this limitation by doing some practical architecture choices.

failed_count_site2=$(s3-check-md5 -versions -access-key minioadmin -secret-key minioadmin -endpoint http://site2-nginx:9002 -bucket testbucket 2>&1 | grep FAILED | wc -l)

## we do not need to fail here, since we are going to test
## upgrading to master, healing and being able to recover
## the last version.
if [ $failed_count_site1 -ne 0 ]; then
	echo "failed with multipart on site1 uploads ${failed_count_site1}"
fi

ERROR: This pool is not scheduled for decommissioning currently.
```

## Canceling a decommission

Stop an on-going decommission in progress, mainly used in situations when the load may be too high and you may want to schedule the decommission at a later point in time.

`mc admin decommission cancel` without an argument, lists out any on-going decommission in progress.

```
λ mc admin decommission cancel alias/

			if ch, ok := val.(chan struct{}); ok && ch != nil {
				// Close previous non-nil channel.
				xioutil.SafeClose(ch)
			}
		}
		globalServiceFreezeCnt = 0 // Don't risk going negative.
	}
	globalServiceFreezeMu.Unlock()