You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
minio/cmd/format-xl.go

760 lines
21 KiB

/*
* Minio Cloud Storage, (C) 2016 Minio, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package cmd
import (
"encoding/json"
"errors"
"fmt"
"reflect"
"sync"
)
const (
// Represents XL backend.
formatBackendXL = "xl"
// formatXLV1.XL.Version
formatXLVersionV1 = "1"
)
// Represents the current backend disk structure
// version under `.minio.sys` and actual data namespace.
// formatXLV1 - structure holds format config version '1'.
type formatXLV1 struct {
formatMetaV1
XL struct {
Version string `json:"version"` // Version of 'xl' format.
Disk string `json:"disk"` // Disk field carries assigned disk uuid.
// JBOD field carries the input disk order generated the first
// time when fresh disks were supplied.
JBOD []string `json:"jbod"`
} `json:"xl"` // XL field holds xl format.
}
/*
All disks online
-----------------
- All Unformatted - format all and return success.
- Some Unformatted - format all and return success.
- Any JBOD inconsistent - return failure // Requires deep inspection, phase2.
- Some are corrupt (missing format.json) - return failure // Requires deep inspection, phase2.
- Any unrecognized disks - return failure
Some disks are offline and we have quorum.
-----------------
- Some unformatted - no heal, return success.
- Any JBOD inconsistent - return failure // Requires deep inspection, phase2.
- Some are corrupt (missing format.json) - return failure // Requires deep inspection, phase2.
- Any unrecognized disks - return failure
No read quorum
-----------------
failure for all cases.
// Pseudo code for managing `format.json`.
// Generic checks.
if (no quorum) return error
if (any disk is corrupt) return error // phase2
if (jbod inconsistent) return error // phase2
if (disks not recognized) // Always error.
// Specific checks.
if (all disks online)
if (all disks return format.json)
if (jbod consistent)
if (all disks recognized)
return
else
if (all disks return format.json not found)
(initialize format)
return
else (some disks return format.json not found)
(heal format)
return
fi
fi
else // No healing at this point forward, some disks are offline or dead.
if (some disks return format.json not found)
if (with force)
// Offline disks are marked as dead.
(heal format) // Offline disks should be marked as dead.
return success
else (without force)
// --force is necessary to heal few drives, because some drives
// are offline. Offline disks will be marked as dead.
return error
fi
fi
*/
// error returned when some disks are offline.
var errSomeDiskOffline = errors.New("some disks are offline")
// errDiskOrderMismatch - returned when disk UUID is not in consistent JBOD order.
var errDiskOrderMismatch = errors.New("disk order mismatch")
// formatErrsSummary - summarizes errors into different classes
func formatErrsSummary(errs []error) (formatCount, unformattedDiskCount,
diskNotFoundCount, corruptedFormatCount, otherErrCount int) {
for _, err := range errs {
switch err {
case errDiskNotFound:
diskNotFoundCount++
case errUnformattedDisk:
unformattedDiskCount++
case errCorruptedFormat:
corruptedFormatCount++
case nil:
// implies that format is not nil
formatCount++
default:
otherErrCount++
}
}
return
}
// loadAllFormats - load all format config from all input disks in parallel.
func loadAllFormats(bootstrapDisks []StorageAPI) ([]*formatXLV1, []error) {
// Initialize sync waitgroup.
var wg = &sync.WaitGroup{}
// Initialize list of errors.
var sErrs = make([]error, len(bootstrapDisks))
// Initialize format configs.
var formats = make([]*formatXLV1, len(bootstrapDisks))
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// Load format from each disk in parallel
for index, disk := range bootstrapDisks {
if disk == nil {
sErrs[index] = errDiskNotFound
continue
}
wg.Add(1)
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// Launch go-routine per disk.
go func(index int, disk StorageAPI) {
defer wg.Done()
format, lErr := loadFormat(disk)
if lErr != nil {
sErrs[index] = lErr
return
}
formats[index] = format
}(index, disk)
}
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// Wait for all go-routines to finish.
wg.Wait()
// Return all formats and nil
return formats, sErrs
}
// genericFormatCheckXL - validates and returns error.
// if (no quorum) return error
// if (any disk is corrupt) return error // phase2
// if (jbod inconsistent) return error // phase2
// if (disks not recognized) // Always error.
func genericFormatCheckXL(formats []*formatXLV1, sErrs []error) (err error) {
// Calculate the errors.
var (
errCorruptFormatCount = 0
errCount = 0
)
// Through all errors calculate the actual errors.
for _, lErr := range sErrs {
if lErr == nil {
continue
}
// These errors are good conditions, means disk is online.
if lErr == errUnformattedDisk || lErr == errVolumeNotFound {
continue
}
if lErr == errCorruptedFormat {
errCorruptFormatCount++
} else {
errCount++
}
}
// Calculate read quorum.
readQuorum := len(formats) / 2
// Validate the err count under read quorum.
if errCount > len(formats)-readQuorum {
return errXLReadQuorum
}
// Check if number of corrupted format under read quorum
if errCorruptFormatCount > len(formats)-readQuorum {
return errCorruptedFormat
}
// Validates if format and JBOD are consistent across all disks.
if err = checkFormatXL(formats); err != nil {
return err
}
// Success..
return nil
}
// isSavedUUIDInOrder - validates if disk uuid is present and valid in all
// available format config JBOD. This function also validates if the disk UUID
// is always available on all JBOD under the same order.
func isSavedUUIDInOrder(uuid string, formats []*formatXLV1) bool {
var orderIndexes []int
// Validate each for format.json for relevant uuid.
for _, format := range formats {
if format == nil {
continue
}
// Validate if UUID is present in JBOD.
uuidIndex := findDiskIndex(uuid, format.XL.JBOD)
if uuidIndex == -1 {
// UUID not found.
errorIf(errDiskNotFound, "Disk %s not found in JBOD list", uuid)
return false
}
// Save the position of UUID present in JBOD.
orderIndexes = append(orderIndexes, uuidIndex+1)
}
// Once uuid is found, verify if the uuid
// present in same order across all format configs.
prevOrderIndex := orderIndexes[0]
for _, orderIndex := range orderIndexes {
if prevOrderIndex != orderIndex {
errorIf(errDiskOrderMismatch, "Disk %s is in wrong order wanted %d, saw %d ", uuid, prevOrderIndex, orderIndex)
return false
}
}
// Returns success, when we have verified if uuid
// is consistent and valid across all format configs.
return true
}
// checkDisksConsistency - checks if all disks are consistent with all JBOD entries on all disks.
func checkDisksConsistency(formats []*formatXLV1) error {
var disks = make([]string, len(formats))
// Collect currently available disk uuids.
for index, format := range formats {
if format == nil {
disks[index] = ""
continue
}
disks[index] = format.XL.Disk
}
// Validate collected uuids and verify JBOD.
for _, uuid := range disks {
if uuid == "" {
continue
}
// Is uuid present on all JBOD ?.
if !isSavedUUIDInOrder(uuid, formats) {
return fmt.Errorf("%s disk not found in JBOD", uuid)
}
}
return nil
}
// checkJBODConsistency - validate xl jbod order if they are consistent.
func checkJBODConsistency(formats []*formatXLV1) error {
var sentinelJBOD []string
// Extract first valid JBOD.
for _, format := range formats {
if format == nil {
continue
}
sentinelJBOD = format.XL.JBOD
break
}
for _, format := range formats {
if format == nil {
continue
}
currentJBOD := format.XL.JBOD
if !reflect.DeepEqual(sentinelJBOD, currentJBOD) {
return errors.New("Inconsistent JBOD found")
}
}
return nil
}
// findDiskIndex returns position of disk in JBOD.
func findDiskIndex(disk string, jbod []string) int {
for index, uuid := range jbod {
if uuid == disk {
return index
}
}
return -1
}
// reorderDisks - reorder disks in JBOD order, and return reference
// format-config. If assignUUIDs is true, it assigns UUIDs to disks
// with missing format configurations in the reference configuration.
func reorderDisks(bootstrapDisks []StorageAPI,
formats []*formatXLV1, assignUUIDs bool) (*formatXLV1,
[]StorageAPI, error) {
// Pick first non-nil format-cfg as reference
var refCfg *formatXLV1
for _, format := range formats {
if format != nil {
refCfg = format
break
}
}
if refCfg == nil {
return nil, nil, fmt.Errorf("could not find any valid config")
}
refJBOD := refCfg.XL.JBOD
// construct reordered disk slice
var newDisks = make([]StorageAPI, len(bootstrapDisks))
for fIndex, format := range formats {
if format == nil {
continue
}
jIndex := findDiskIndex(format.XL.Disk, refJBOD)
if jIndex == -1 {
return nil, nil, errors.New("Unrecognized uuid " + format.XL.Disk + " found")
}
newDisks[jIndex] = bootstrapDisks[fIndex]
}
if assignUUIDs {
// Based on orderedDisks generate new UUIDs in the ref. config
// for disks without format-configs.
for index, disk := range newDisks {
if disk == nil {
refCfg.XL.JBOD[index] = mustGetUUID()
}
}
}
return refCfg, newDisks, nil
}
// loadFormat - loads format.json from disk.
func loadFormat(disk StorageAPI) (format *formatXLV1, err error) {
buf, err := disk.ReadAll(minioMetaBucket, formatConfigFile)
if err != nil {
// 'file not found' and 'volume not found' as
// same. 'volume not found' usually means its a fresh disk.
if err == errFileNotFound || err == errVolumeNotFound {
var vols []VolInfo
vols, err = disk.ListVols()
if err != nil {
return nil, err
}
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
if len(vols) > 1 || (len(vols) == 1 &&
vols[0].Name != minioMetaBucket) {
// 'format.json' not found, but we
// found user data.
return nil, errCorruptedFormat
}
// No other data found, its a fresh disk.
return nil, errUnformattedDisk
}
return nil, err
}
// Try to decode format json into formatConfigV1 struct.
format = &formatXLV1{}
if err = json.Unmarshal(buf, format); err != nil {
return nil, err
}
// Success.
return format, nil
}
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// collectNSaveNewFormatConfigs - generates new format configs based on
// the given ref. config and saves on each disk
func collectNSaveNewFormatConfigs(referenceConfig *formatXLV1,
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
orderedDisks []StorageAPI, dryRun bool) error {
// Collect new format configs that need to be written.
var newFormatConfigs = make([]*formatXLV1, len(orderedDisks))
for index := range orderedDisks {
// New configs are generated since we are going
// to re-populate across all disks.
config := &formatXLV1{}
config.Version = referenceConfig.Version
config.Format = referenceConfig.Format
config.XL.Version = referenceConfig.XL.Version
config.XL.Disk = referenceConfig.XL.JBOD[index]
config.XL.JBOD = referenceConfig.XL.JBOD
newFormatConfigs[index] = config
}
// Initialize meta volume, if volume already exists ignores it.
if err := initMetaVolume(orderedDisks); err != nil {
return fmt.Errorf("Unable to initialize '.minio.sys' meta volume, %s", err)
}
// Save new `format.json` across all disks, in JBOD order.
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
if !dryRun {
return saveFormatXL(orderedDisks, newFormatConfigs)
}
return nil
}
// Heals any missing format.json on the drives. Returns error only for
// unexpected errors as regular errors can be ignored since there
// might be enough quorum to be operational. Heals only fresh disks.
func healFormatXLFreshDisks(storageDisks []StorageAPI,
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
formats []*formatXLV1, dryRun bool) error {
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// Reorder disks based on JBOD order, and get reference config.
referenceConfig, orderedDisks, err := reorderDisks(storageDisks,
formats, true)
if err != nil {
return err
}
// Fill in the missing disk back from format configs.
// We need to make sure we have kept the previous order
// and allowed fresh disks to be arranged anywhere.
// Following block facilitates to put fresh disks.
for index, format := range formats {
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
if format != nil {
continue
}
// Format is missing so we go through ordered disks.
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// At this point when disk is missing the fresh disk
// in the stack get it back from storageDisks.
for oIndex, disk := range orderedDisks {
if disk == nil {
orderedDisks[oIndex] = storageDisks[index]
break
}
}
}
// apply new format config and save to all disks
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
return collectNSaveNewFormatConfigs(referenceConfig, orderedDisks,
dryRun)
}
// collectUnAssignedDisks - collect disks unassigned to orderedDisks
// from storageDisks and return them.
func collectUnAssignedDisks(storageDisks, orderedDisks []StorageAPI) (
uDisks []StorageAPI) {
// search for each disk from storageDisks in orderedDisks
for i := range storageDisks {
found := false
for j := range orderedDisks {
if storageDisks[i] == orderedDisks[j] {
found = true
break
}
}
if !found {
// append not found disk to result
uDisks = append(uDisks, storageDisks[i])
}
}
return uDisks
}
// Inspect the content of all disks to guess the right order according
// to the format files. The right order is represented in orderedDisks
func reorderDisksByInspection(orderedDisks, storageDisks []StorageAPI,
formats []*formatXLV1) ([]StorageAPI, error) {
for index, format := range formats {
if format != nil {
continue
}
vols, err := storageDisks[index].ListVols()
if err != nil {
return nil, err
}
if len(vols) == 0 {
continue
}
volName := ""
// Avoid picking minioMetaBucket because ListVols()
// returns a non ordered list
for i := range vols {
if vols[i].Name != minioMetaBucket {
volName = vols[i].Name
break
}
}
if volName == "" {
continue
}
objects, err := storageDisks[index].ListDir(volName, "")
if err != nil {
return nil, err
}
if len(objects) == 0 {
continue
}
xlData, err := readXLMeta(storageDisks[index], volName, objects[0])
if err != nil {
if err == errFileNotFound {
continue
}
return nil, err
}
diskIndex := -1
for i, d := range xlData.Erasure.Distribution {
if d == xlData.Erasure.Index {
diskIndex = i
}
}
// Check for found results
if diskIndex == -1 || orderedDisks[diskIndex] != nil {
// Some inconsistent data are found, exit immediately.
return nil, errCorruptedFormat
}
orderedDisks[diskIndex] = storageDisks[index]
}
return orderedDisks, nil
}
// Heals corrupted format json in all disks
func healFormatXLCorruptedDisks(storageDisks []StorageAPI,
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
formats []*formatXLV1, dryRun bool) error {
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// Reorder disks based on JBOD order, and update ref. config.
referenceConfig, orderedDisks, err := reorderDisks(storageDisks,
formats, true)
if err != nil {
return err
}
// For disks with corrupted formats, inspect the disks
// contents to guess the disks order
orderedDisks, err = reorderDisksByInspection(orderedDisks, storageDisks,
formats)
if err != nil {
return err
}
// At this stage, all disks with corrupted formats but with
// objects inside found their way. Now take care of
// unformatted disks, which are the `unAssignedDisks`
unAssignedDisks := collectUnAssignedDisks(storageDisks, orderedDisks)
// Assign unassigned disks to nil elements in orderedDisks
for i, disk := range orderedDisks {
if disk == nil && len(unAssignedDisks) > 0 {
orderedDisks[i] = unAssignedDisks[0]
unAssignedDisks = unAssignedDisks[1:]
}
}
Move admin APIs to new path and add redesigned heal APIs (#5351) - Changes related to moving admin APIs - admin APIs now have an endpoint under /minio/admin - admin APIs are now versioned - a new API to server the version is added at "GET /minio/admin/version" and all API operations have the path prefix /minio/admin/v1/<operation> - new service stop API added - credentials change API is moved to /minio/admin/v1/config/credential - credentials change API and configuration get/set API now require TLS so that credentials are protected - all API requests now receive JSON - heal APIs are disabled as they will be changed substantially - Heal API changes Heal API is now provided at a single endpoint with the ability for a client to start a heal sequence on all the data in the server, a single bucket, or under a prefix within a bucket. When a heal sequence is started, the server returns a unique token that needs to be used for subsequent 'status' requests to fetch heal results. On each status request from the client, the server returns heal result records that it has accumulated since the previous status request. The server accumulates upto 1000 records and pauses healing further objects until the client requests for status. If the client does not request any further records for a long time, the server aborts the heal sequence automatically. A heal result record is returned for each entity healed on the server, such as system metadata, object metadata, buckets and objects, and has information about the before and after states on each disk. A client may request to force restart a heal sequence - this causes the running heal sequence to be aborted at the next safe spot and starts a new heal sequence.
7 years ago
// generate and write new configs to all disks
return collectNSaveNewFormatConfigs(referenceConfig, orderedDisks,
dryRun)
}
// loadFormatXL - loads XL `format.json` and returns back properly
// ordered storage slice based on `format.json`.
func loadFormatXL(bootstrapDisks []StorageAPI, readQuorum int) (disks []StorageAPI, err error) {
var unformattedDisksFoundCnt = 0
var diskNotFoundCount = 0
var corruptedDisksFoundCnt = 0
formats := make([]*formatXLV1, len(bootstrapDisks))
// Try to load `format.json` bootstrap disks.
for index, disk := range bootstrapDisks {
if disk == nil {
diskNotFoundCount++
continue
}
var formatXL *formatXLV1
formatXL, err = loadFormat(disk)
if err != nil {
if err == errUnformattedDisk {
unformattedDisksFoundCnt++
continue
} else if err == errDiskNotFound {
diskNotFoundCount++
continue
} else if err == errCorruptedFormat {
corruptedDisksFoundCnt++
continue
}
return nil, err
}
// Save valid formats.
formats[index] = formatXL
}
// If all disks indicate that 'format.json' is not available return 'errUnformattedDisk'.
if unformattedDisksFoundCnt > len(bootstrapDisks)-readQuorum {
return nil, errUnformattedDisk
} else if corruptedDisksFoundCnt > len(bootstrapDisks)-readQuorum {
return nil, errCorruptedFormat
} else if diskNotFoundCount == len(bootstrapDisks) {
return nil, errDiskNotFound
} else if diskNotFoundCount > len(bootstrapDisks)-readQuorum {
return nil, errXLReadQuorum
}
// Validate the format configs read are correct.
if err = checkFormatXL(formats); err != nil {
return nil, err
}
// Erasure code requires disks to be presented in the same
// order each time.
_, orderedDisks, err := reorderDisks(bootstrapDisks, formats,
false)
return orderedDisks, err
}
func checkFormatXLValue(formatXL *formatXLV1) error {
// Validate format version and format type.
if formatXL.Version != formatMetaVersionV1 {
return fmt.Errorf("Unsupported version of backend format [%s] found", formatXL.Version)
}
if formatXL.Format != formatBackendXL {
return fmt.Errorf("Unsupported backend format [%s] found", formatXL.Format)
}
if formatXL.XL.Version != formatXLVersionV1 {
return fmt.Errorf("Unsupported XL backend format found [%s]", formatXL.XL.Version)
}
return nil
}
func checkFormatXLValues(formats []*formatXLV1) (int, error) {
for i, formatXL := range formats {
if formatXL == nil {
continue
}
if err := checkFormatXLValue(formatXL); err != nil {
return i, err
}
if len(formats) != len(formatXL.XL.JBOD) {
return i, fmt.Errorf("Number of disks %d did not match the backend format %d",
len(formats), len(formatXL.XL.JBOD))
}
}
return -1, nil
}
// checkFormatXL - verifies if format.json format is intact.
func checkFormatXL(formats []*formatXLV1) error {
if _, err := checkFormatXLValues(formats); err != nil {
return err
}
if err := checkJBODConsistency(formats); err != nil {
return err
}
return checkDisksConsistency(formats)
}
// saveFormatXL - populates `format.json` on disks in its order.
func saveFormatXL(storageDisks []StorageAPI, formats []*formatXLV1) error {
var errs = make([]error, len(storageDisks))
var wg = &sync.WaitGroup{}
// Write `format.json` to all disks.
for index, disk := range storageDisks {
if disk == nil {
continue
}
wg.Add(1)
go func(index int, disk StorageAPI, format *formatXLV1) {
defer wg.Done()
// Marshal and write to disk.
formatBytes, err := json.Marshal(format)
if err != nil {
errs[index] = err
return
}
// Purge any existing temporary file, okay to ignore errors here.
disk.DeleteFile(minioMetaBucket, formatConfigFileTmp)
// Append file `format.json.tmp`.
if err = disk.AppendFile(minioMetaBucket, formatConfigFileTmp, formatBytes); err != nil {
errs[index] = err
return
}
// Rename file `format.json.tmp` --> `format.json`.
if err = disk.RenameFile(minioMetaBucket, formatConfigFileTmp, minioMetaBucket, formatConfigFile); err != nil {
errs[index] = err
return
}
}(index, disk, formats[index])
}
// Wait for the routines to finish.
wg.Wait()
// Validate if we encountered any errors, return quickly.
for _, err := range errs {
if err != nil {
// Failure.
return err
}
}
// Success.
return nil
}
// Return a slice of format, to be used to format uninitialized disks.
func newFormatXLV1(diskCount int) []*formatXLV1 {
var jbod = make([]string, diskCount)
var formats = make([]*formatXLV1, diskCount)
for i := 0; i < diskCount; i++ {
format := &formatXLV1{}
format.Version = formatMetaVersionV1
format.Format = formatBackendXL
format.XL.Version = formatXLVersionV1
format.XL.Disk = mustGetUUID()
formats[i] = format
jbod[i] = formats[i].XL.Disk
}
// Update the jbod entries.
for i := 0; i < diskCount; i++ {
formats[i].XL.JBOD = jbod
}
return formats
}
// initFormatXL - save XL format configuration on all disks.
func initFormatXL(storageDisks []StorageAPI) (err error) {
// Initialize meta volume, if volume already exists ignores it.
if err := initMetaVolume(storageDisks); err != nil {
return fmt.Errorf("Unable to initialize '.minio.sys' meta volume, %s", err)
}
// Save formats `format.json` across all disks.
return saveFormatXL(storageDisks, newFormatXLV1(len(storageDisks)))
}