1 // Copyright (C) The Arvados Authors. All rights reserved.
3 // SPDX-License-Identifier: AGPL-3.0
19 "git.curoverse.com/arvados.git/sdk/go/arvados"
20 "git.curoverse.com/arvados.git/sdk/go/keepclient"
21 "github.com/sirupsen/logrus"
24 // Balancer compares the contents of keepstore servers with the
25 // collections stored in Arvados, and issues pull/trash requests
26 // needed to get (closer to) the optimal data layout.
28 // In the optimal data layout: every data block referenced by a
29 // collection is replicated at least as many times as desired by the
30 // collection; there are no unreferenced data blocks older than
31 // BlobSignatureTTL; and all N existing replicas of a given data block
32 // are in the N best positions in rendezvous probe order.
33 type Balancer struct {
39 KeepServices map[string]*KeepService
40 DefaultReplication int
45 mountsByClass map[string]map[*KeepMount]bool
47 serviceRoots map[string]string
53 // Run performs a balance operation using the given config and
54 // runOptions, and returns RunOptions suitable for passing to a
55 // subsequent balance operation.
57 // Run should only be called once on a given Balancer object.
61 // runOptions, err = (&Balancer{}).Run(config, runOptions)
62 func (bal *Balancer) Run(config Config, runOptions RunOptions) (nextRunOptions RunOptions, err error) {
63 nextRunOptions = runOptions
65 defer bal.time("sweep", "wall clock time to run one full sweep")()
67 if len(config.KeepServiceList.Items) > 0 {
68 err = bal.SetKeepServices(config.KeepServiceList)
70 err = bal.DiscoverKeepServices(&config.Client, config.KeepServiceTypes)
76 for _, srv := range bal.KeepServices {
77 err = srv.discoverMounts(&config.Client)
84 if err = bal.CheckSanityEarly(&config.Client); err != nil {
87 rs := bal.rendezvousState()
88 if runOptions.CommitTrash && rs != runOptions.SafeRendezvousState {
89 if runOptions.SafeRendezvousState != "" {
90 bal.logf("notice: KeepServices list has changed since last run")
92 bal.logf("clearing existing trash lists, in case the new rendezvous order differs from previous run")
93 if err = bal.ClearTrashLists(&config.Client); err != nil {
96 // The current rendezvous state becomes "safe" (i.e.,
97 // OK to compute changes for that state without
98 // clearing existing trash lists) only now, after we
99 // succeed in clearing existing trash lists.
100 nextRunOptions.SafeRendezvousState = rs
102 if err = bal.GetCurrentState(&config.Client, config.CollectionBatchSize, config.CollectionBuffers); err != nil {
105 bal.ComputeChangeSets()
106 bal.PrintStatistics()
107 if err = bal.CheckSanityLate(); err != nil {
110 if runOptions.CommitPulls {
111 err = bal.CommitPulls(&config.Client)
113 // Skip trash if we can't pull. (Too cautious?)
117 if runOptions.CommitTrash {
118 err = bal.CommitTrash(&config.Client)
123 // SetKeepServices sets the list of KeepServices to operate on.
124 func (bal *Balancer) SetKeepServices(srvList arvados.KeepServiceList) error {
125 bal.KeepServices = make(map[string]*KeepService)
126 for _, srv := range srvList.Items {
127 bal.KeepServices[srv.UUID] = &KeepService{
129 ChangeSet: &ChangeSet{},
135 // DiscoverKeepServices sets the list of KeepServices by calling the
136 // API to get a list of all services, and selecting the ones whose
137 // ServiceType is in okTypes.
138 func (bal *Balancer) DiscoverKeepServices(c *arvados.Client, okTypes []string) error {
139 bal.KeepServices = make(map[string]*KeepService)
140 ok := make(map[string]bool)
141 for _, t := range okTypes {
144 return c.EachKeepService(func(srv arvados.KeepService) error {
145 if ok[srv.ServiceType] {
146 bal.KeepServices[srv.UUID] = &KeepService{
148 ChangeSet: &ChangeSet{},
151 bal.logf("skipping %v with service type %q", srv.UUID, srv.ServiceType)
157 func (bal *Balancer) cleanupMounts() {
158 rwdev := map[string]*KeepService{}
159 for _, srv := range bal.KeepServices {
160 for _, mnt := range srv.mounts {
161 if !mnt.ReadOnly && mnt.DeviceID != "" {
162 rwdev[mnt.DeviceID] = srv
166 // Drop the readonly mounts whose device is mounted RW
168 for _, srv := range bal.KeepServices {
169 var dedup []*KeepMount
170 for _, mnt := range srv.mounts {
171 if mnt.ReadOnly && rwdev[mnt.DeviceID] != nil {
172 bal.logf("skipping srv %s readonly mount %q because same device %q is mounted read-write on srv %s", srv, mnt.UUID, mnt.DeviceID, rwdev[mnt.DeviceID])
174 dedup = append(dedup, mnt)
179 for _, srv := range bal.KeepServices {
180 for _, mnt := range srv.mounts {
181 if mnt.Replication <= 0 {
182 log.Printf("%s: mount %s reports replication=%d, using replication=1", srv, mnt.UUID, mnt.Replication)
189 // CheckSanityEarly checks for configuration and runtime errors that
190 // can be detected before GetCurrentState() and ComputeChangeSets()
193 // If it returns an error, it is pointless to run GetCurrentState or
194 // ComputeChangeSets: after doing so, the statistics would be
195 // meaningless and it would be dangerous to run any Commit methods.
196 func (bal *Balancer) CheckSanityEarly(c *arvados.Client) error {
197 u, err := c.CurrentUser()
199 return fmt.Errorf("CurrentUser(): %v", err)
201 if !u.IsActive || !u.IsAdmin {
202 return fmt.Errorf("current user (%s) is not an active admin user", u.UUID)
204 for _, srv := range bal.KeepServices {
205 if srv.ServiceType == "proxy" {
206 return fmt.Errorf("config error: %s: proxy servers cannot be balanced", srv)
212 // rendezvousState returns a fingerprint (e.g., a sorted list of
213 // UUID+host+port) of the current set of keep services.
214 func (bal *Balancer) rendezvousState() string {
215 srvs := make([]string, 0, len(bal.KeepServices))
216 for _, srv := range bal.KeepServices {
217 srvs = append(srvs, srv.String())
220 return strings.Join(srvs, "; ")
223 // ClearTrashLists sends an empty trash list to each keep
224 // service. Calling this before GetCurrentState avoids races.
226 // When a block appears in an index, we assume that replica will still
227 // exist after we delete other replicas on other servers. However,
228 // it's possible that a previous rebalancing operation made different
229 // decisions (e.g., servers were added/removed, and rendezvous order
230 // changed). In this case, the replica might already be on that
231 // server's trash list, and it might be deleted before we send a
232 // replacement trash list.
234 // We avoid this problem if we clear all trash lists before getting
235 // indexes. (We also assume there is only one rebalancing process
236 // running at a time.)
237 func (bal *Balancer) ClearTrashLists(c *arvados.Client) error {
238 for _, srv := range bal.KeepServices {
239 srv.ChangeSet = &ChangeSet{}
241 return bal.CommitTrash(c)
244 // GetCurrentState determines the current replication state, and the
245 // desired replication level, for every block that is either
246 // retrievable or referenced.
248 // It determines the current replication state by reading the block index
249 // from every known Keep service.
251 // It determines the desired replication level by retrieving all
252 // collection manifests in the database (API server).
254 // It encodes the resulting information in BlockStateMap.
255 func (bal *Balancer) GetCurrentState(c *arvados.Client, pageSize, bufs int) error {
256 defer bal.time("get_state", "wall clock time to get current state")()
257 bal.BlockStateMap = NewBlockStateMap()
259 dd, err := c.DiscoveryDocument()
263 bal.DefaultReplication = dd.DefaultCollectionReplication
264 bal.MinMtime = time.Now().UnixNano() - dd.BlobSignatureTTL*1e9
266 errs := make(chan error, 1)
267 wg := sync.WaitGroup{}
269 // When a device is mounted more than once, we will get its
270 // index only once, and call AddReplicas on all of the mounts.
271 // equivMount keys are the mounts that will be indexed, and
272 // each value is a list of mounts to apply the received index
274 equivMount := map[*KeepMount][]*KeepMount{}
275 // deviceMount maps each device ID to the one mount that will
276 // be indexed for that device.
277 deviceMount := map[string]*KeepMount{}
278 for _, srv := range bal.KeepServices {
279 for _, mnt := range srv.mounts {
280 equiv := deviceMount[mnt.DeviceID]
283 if mnt.DeviceID != "" {
284 deviceMount[mnt.DeviceID] = equiv
287 equivMount[equiv] = append(equivMount[equiv], mnt)
291 // Start one goroutine for each (non-redundant) mount:
292 // retrieve the index, and add the returned blocks to
294 for _, mounts := range equivMount {
296 go func(mounts []*KeepMount) {
298 bal.logf("mount %s: retrieve index from %s", mounts[0], mounts[0].KeepService)
299 idx, err := mounts[0].KeepService.IndexMount(c, mounts[0].UUID, "")
302 case errs <- fmt.Errorf("%s: retrieve index: %v", mounts[0], err):
308 // Some other goroutine encountered an
309 // error -- any further effort here
313 for _, mount := range mounts {
314 bal.logf("%s: add %d entries to map", mount, len(idx))
315 bal.BlockStateMap.AddReplicas(mount, idx)
316 bal.logf("%s: added %d entries to map at %dx (%d replicas)", mount, len(idx), mount.Replication, len(idx)*mount.Replication)
318 bal.logf("mount %s: index done", mounts[0])
322 // collQ buffers incoming collections so we can start fetching
323 // the next page without waiting for the current page to
324 // finish processing.
325 collQ := make(chan arvados.Collection, bufs)
327 // Start a goroutine to process collections. (We could use a
328 // worker pool here, but even with a single worker we already
329 // process collections much faster than we can retrieve them.)
333 for coll := range collQ {
334 err := bal.addCollection(coll)
348 // Start a goroutine to retrieve all collections from the
349 // Arvados database and send them to collQ for processing.
353 err = EachCollection(c, pageSize,
354 func(coll arvados.Collection) error {
357 // some other GetCurrentState
358 // error happened: no point
361 return fmt.Errorf("")
364 }, func(done, total int) {
365 bal.logf("collections: %d/%d", done, total)
383 func (bal *Balancer) addCollection(coll arvados.Collection) error {
384 blkids, err := coll.SizedDigests()
387 bal.errors = append(bal.errors, fmt.Errorf("%v: %v", coll.UUID, err))
391 repl := bal.DefaultReplication
392 if coll.ReplicationDesired != nil {
393 repl = *coll.ReplicationDesired
395 debugf("%v: %d block x%d", coll.UUID, len(blkids), repl)
396 bal.BlockStateMap.IncreaseDesired(coll.StorageClassesDesired, repl, blkids)
400 // ComputeChangeSets compares, for each known block, the current and
401 // desired replication states. If it is possible to get closer to the
402 // desired state by copying or deleting blocks, it adds those changes
403 // to the relevant KeepServices' ChangeSets.
405 // It does not actually apply any of the computed changes.
406 func (bal *Balancer) ComputeChangeSets() {
407 // This just calls balanceBlock() once for each block, using a
408 // pool of worker goroutines.
409 defer bal.time("changeset_compute", "wall clock time to compute changesets")()
410 bal.setupLookupTables()
412 type balanceTask struct {
413 blkid arvados.SizedDigest
416 workers := runtime.GOMAXPROCS(-1)
417 todo := make(chan balanceTask, workers)
419 bal.BlockStateMap.Apply(func(blkid arvados.SizedDigest, blk *BlockState) {
427 results := make(chan balanceResult, workers)
429 var wg sync.WaitGroup
430 for i := 0; i < workers; i++ {
433 for work := range todo {
434 results <- bal.balanceBlock(work.blkid, work.blk)
442 bal.collectStatistics(results)
445 func (bal *Balancer) setupLookupTables() {
446 bal.serviceRoots = make(map[string]string)
447 bal.classes = []string{"default"}
448 bal.mountsByClass = map[string]map[*KeepMount]bool{"default": {}}
450 for _, srv := range bal.KeepServices {
451 bal.serviceRoots[srv.UUID] = srv.UUID
452 for _, mnt := range srv.mounts {
455 // All mounts on a read-only service are
456 // effectively read-only.
457 mnt.ReadOnly = mnt.ReadOnly || srv.ReadOnly
459 if len(mnt.StorageClasses) == 0 {
460 bal.mountsByClass["default"][mnt] = true
463 for _, class := range mnt.StorageClasses {
464 if mbc := bal.mountsByClass[class]; mbc == nil {
465 bal.classes = append(bal.classes, class)
466 bal.mountsByClass[class] = map[*KeepMount]bool{mnt: true}
473 // Consider classes in lexicographic order to avoid flapping
474 // between balancing runs. The outcome of the "prefer a mount
475 // we're already planning to use for a different storage
476 // class" case in balanceBlock depends on the order classes
478 sort.Strings(bal.classes)
488 var changeName = map[int]string{
491 changeTrash: "trash",
495 type balanceResult struct {
497 blkid arvados.SizedDigest
500 classState map[string]balancedBlockState
503 // balanceBlock compares current state to desired state for a single
504 // block, and makes the appropriate ChangeSet calls.
505 func (bal *Balancer) balanceBlock(blkid arvados.SizedDigest, blk *BlockState) balanceResult {
506 debugf("balanceBlock: %v %+v", blkid, blk)
509 mnt *KeepMount // never nil
510 repl *Replica // replica already stored here (or nil)
511 want bool // we should pull/leave a replica here
514 // Build a list of all slots (one per mounted volume).
515 slots := make([]slot, 0, bal.mounts)
516 for _, srv := range bal.KeepServices {
517 for _, mnt := range srv.mounts {
519 for r := range blk.Replicas {
520 if blk.Replicas[r].KeepMount == mnt {
521 repl = &blk.Replicas[r]
524 // Initial value of "want" is "have, and can't
525 // delete". These untrashable replicas get
526 // prioritized when sorting slots: otherwise,
527 // non-optimal readonly copies would cause us
529 slots = append(slots, slot{
532 want: repl != nil && mnt.ReadOnly,
537 uuids := keepclient.NewRootSorter(bal.serviceRoots, string(blkid[:32])).GetSortedRoots()
538 srvRendezvous := make(map[*KeepService]int, len(uuids))
539 for i, uuid := range uuids {
540 srv := bal.KeepServices[uuid]
541 srvRendezvous[srv] = i
544 // Below we set underreplicated=true if we find any storage
545 // class that's currently underreplicated -- in that case we
546 // won't want to trash any replicas.
547 underreplicated := false
549 classState := make(map[string]balancedBlockState, len(bal.classes))
550 unsafeToDelete := make(map[int64]bool, len(slots))
551 for _, class := range bal.classes {
552 desired := blk.Desired[class]
554 countedDev := map[string]bool{}
556 for _, slot := range slots {
557 if slot.repl != nil && bal.mountsByClass[class][slot.mnt] && !countedDev[slot.mnt.DeviceID] {
558 have += slot.mnt.Replication
559 if slot.mnt.DeviceID != "" {
560 countedDev[slot.mnt.DeviceID] = true
564 classState[class] = balancedBlockState{
566 surplus: have - desired,
573 // Sort the slots by desirability.
574 sort.Slice(slots, func(i, j int) bool {
575 si, sj := slots[i], slots[j]
576 if classi, classj := bal.mountsByClass[class][si.mnt], bal.mountsByClass[class][sj.mnt]; classi != classj {
577 // Prefer a mount that satisfies the
579 return bal.mountsByClass[class][si.mnt]
580 } else if si.want != sj.want {
581 // Prefer a mount that will have a
582 // replica no matter what we do here
583 // -- either because it already has an
584 // untrashable replica, or because we
585 // already need it to satisfy a
586 // different storage class.
588 } else if orderi, orderj := srvRendezvous[si.mnt.KeepService], srvRendezvous[sj.mnt.KeepService]; orderi != orderj {
589 // Prefer a better rendezvous
591 return orderi < orderj
592 } else if repli, replj := si.repl != nil, sj.repl != nil; repli != replj {
593 // Prefer a mount that already has a
597 // If pull/trash turns out to be
598 // needed, distribute the
599 // new/remaining replicas uniformly
600 // across qualifying mounts on a given
602 return rendezvousLess(si.mnt.DeviceID, sj.mnt.DeviceID, blkid)
606 // Servers/mounts/devices (with or without existing
607 // replicas) that are part of the best achievable
608 // layout for this storage class.
609 wantSrv := map[*KeepService]bool{}
610 wantMnt := map[*KeepMount]bool{}
611 wantDev := map[string]bool{}
612 // Positions (with existing replicas) that have been
613 // protected (via unsafeToDelete) to ensure we don't
614 // reduce replication below desired level when
615 // trashing replicas that aren't optimal positions for
616 // any storage class.
617 protMnt := map[*KeepMount]bool{}
618 // Replication planned so far (corresponds to wantMnt).
620 // Protected replication (corresponds to protMnt).
623 // trySlot tries using a slot to meet requirements,
624 // and returns true if all requirements are met.
625 trySlot := func(i int) bool {
627 if wantMnt[slot.mnt] || wantDev[slot.mnt.DeviceID] {
628 // Already allocated a replica to this
629 // backend device, possibly on a
633 if replProt < desired && slot.repl != nil && !protMnt[slot.mnt] {
634 unsafeToDelete[slot.repl.Mtime] = true
635 protMnt[slot.mnt] = true
636 replProt += slot.mnt.Replication
638 if replWant < desired && (slot.repl != nil || !slot.mnt.ReadOnly) {
640 wantSrv[slot.mnt.KeepService] = true
641 wantMnt[slot.mnt] = true
642 if slot.mnt.DeviceID != "" {
643 wantDev[slot.mnt.DeviceID] = true
645 replWant += slot.mnt.Replication
647 return replProt >= desired && replWant >= desired
650 // First try to achieve desired replication without
651 // using the same server twice.
653 for i := 0; i < len(slots) && !done; i++ {
654 if !wantSrv[slots[i].mnt.KeepService] {
659 // If that didn't suffice, do another pass without the
660 // "distinct services" restriction. (Achieving the
661 // desired volume replication on fewer than the
662 // desired number of services is better than
663 // underreplicating.)
664 for i := 0; i < len(slots) && !done; i++ {
668 if !underreplicated {
670 for _, slot := range slots {
671 if slot.repl == nil || !bal.mountsByClass[class][slot.mnt] {
674 if safe += slot.mnt.Replication; safe >= desired {
678 underreplicated = safe < desired
681 // set the unachievable flag if there aren't enough
682 // slots offering the relevant storage class. (This is
683 // as easy as checking slots[desired] because we
684 // already sorted the qualifying slots to the front.)
685 if desired >= len(slots) || !bal.mountsByClass[class][slots[desired].mnt] {
686 cs := classState[class]
687 cs.unachievable = true
688 classState[class] = cs
691 // Avoid deleting wanted replicas from devices that
692 // are mounted on multiple servers -- even if they
693 // haven't already been added to unsafeToDelete
694 // because the servers report different Mtimes.
695 for _, slot := range slots {
696 if slot.repl != nil && wantDev[slot.mnt.DeviceID] {
697 unsafeToDelete[slot.repl.Mtime] = true
702 // TODO: If multiple replicas are trashable, prefer the oldest
703 // replica that doesn't have a timestamp collision with
706 countedDev := map[string]bool{}
708 for _, slot := range slots {
709 if countedDev[slot.mnt.DeviceID] {
713 want += slot.mnt.Replication
715 if slot.repl != nil {
716 have += slot.mnt.Replication
718 if slot.mnt.DeviceID != "" {
719 countedDev[slot.mnt.DeviceID] = true
724 for _, slot := range slots {
725 // TODO: request a Touch if Mtime is duplicated.
728 case !underreplicated && !slot.want && slot.repl != nil && slot.repl.Mtime < bal.MinMtime && !unsafeToDelete[slot.repl.Mtime]:
729 slot.mnt.KeepService.AddTrash(Trash{
731 Mtime: slot.repl.Mtime,
735 case len(blk.Replicas) == 0:
737 case slot.repl == nil && slot.want && !slot.mnt.ReadOnly:
738 slot.mnt.KeepService.AddPull(Pull{
740 From: blk.Replicas[0].KeepMount.KeepService,
747 if bal.Dumper != nil {
749 if slot.repl != nil {
750 mtime = slot.repl.Mtime
752 srv := slot.mnt.KeepService
753 changes = append(changes, fmt.Sprintf("%s:%d/%s=%s,%d", srv.ServiceHost, srv.ServicePort, slot.mnt.UUID, changeName[change], mtime))
756 if bal.Dumper != nil {
757 bal.Dumper.Printf("%s have=%d want=%v %s", blkid, have, want, strings.Join(changes, " "))
759 return balanceResult{
764 classState: classState,
768 type blocksNBytes struct {
774 func (bb blocksNBytes) String() string {
775 return fmt.Sprintf("%d replicas (%d blocks, %d bytes)", bb.replicas, bb.blocks, bb.bytes)
778 type balancerStats struct {
783 underrep blocksNBytes
784 unachievable blocksNBytes
785 justright blocksNBytes
791 classStats map[string]replicationStats
793 // collectionBytes / collectionBlockBytes = deduplication ratio
794 collectionBytes int64 // sum(bytes in referenced blocks) across all collections
795 collectionBlockBytes int64 // sum(block size) across all blocks referenced by collections
796 collectionBlockRefs int64 // sum(number of blocks referenced) across all collections
797 collectionBlocks int64 // number of blocks referenced by any collection
800 func (s *balancerStats) dedupByteRatio() float64 {
801 if s.collectionBlockBytes == 0 {
804 return float64(s.collectionBytes) / float64(s.collectionBlockBytes)
807 func (s *balancerStats) dedupBlockRatio() float64 {
808 if s.collectionBlocks == 0 {
811 return float64(s.collectionBlockRefs) / float64(s.collectionBlocks)
814 type replicationStats struct {
818 unachievable blocksNBytes
821 type balancedBlockState struct {
827 func (bal *Balancer) collectStatistics(results <-chan balanceResult) {
829 s.replHistogram = make([]int, 2)
830 s.classStats = make(map[string]replicationStats, len(bal.classes))
831 for result := range results {
832 surplus := result.have - result.want
833 bytes := result.blkid.Size()
835 if rc := int64(result.blk.RefCount); rc > 0 {
836 s.collectionBytes += rc * bytes
837 s.collectionBlockBytes += bytes
838 s.collectionBlockRefs += rc
842 for class, state := range result.classState {
843 cs := s.classStats[class]
844 if state.unachievable {
845 cs.unachievable.blocks++
846 cs.unachievable.bytes += bytes
848 if state.desired > 0 {
849 cs.desired.replicas += state.desired
851 cs.desired.bytes += bytes * int64(state.desired)
853 if state.surplus > 0 {
854 cs.surplus.replicas += state.surplus
856 cs.surplus.bytes += bytes * int64(state.surplus)
857 } else if state.surplus < 0 {
858 cs.short.replicas += -state.surplus
860 cs.short.bytes += bytes * int64(-state.surplus)
862 s.classStats[class] = cs
866 case result.have == 0 && result.want > 0:
867 s.lost.replicas -= surplus
869 s.lost.bytes += bytes * int64(-surplus)
871 s.underrep.replicas -= surplus
873 s.underrep.bytes += bytes * int64(-surplus)
874 case surplus > 0 && result.want == 0:
875 counter := &s.garbage
876 for _, r := range result.blk.Replicas {
877 if r.Mtime >= bal.MinMtime {
882 counter.replicas += surplus
884 counter.bytes += bytes * int64(surplus)
886 s.overrep.replicas += surplus
888 s.overrep.bytes += bytes * int64(result.have-result.want)
890 s.justright.replicas += result.want
892 s.justright.bytes += bytes * int64(result.want)
896 s.desired.replicas += result.want
898 s.desired.bytes += bytes * int64(result.want)
901 s.current.replicas += result.have
903 s.current.bytes += bytes * int64(result.have)
906 for len(s.replHistogram) <= result.have {
907 s.replHistogram = append(s.replHistogram, 0)
909 s.replHistogram[result.have]++
911 for _, srv := range bal.KeepServices {
912 s.pulls += len(srv.ChangeSet.Pulls)
913 s.trashes += len(srv.ChangeSet.Trashes)
916 bal.Metrics.UpdateStats(s)
919 // PrintStatistics writes statistics about the computed changes to
920 // bal.Logger. It should not be called until ComputeChangeSets has
922 func (bal *Balancer) PrintStatistics() {
924 bal.logf("%s lost (0=have<want)", bal.stats.lost)
925 bal.logf("%s underreplicated (0<have<want)", bal.stats.underrep)
926 bal.logf("%s just right (have=want)", bal.stats.justright)
927 bal.logf("%s overreplicated (have>want>0)", bal.stats.overrep)
928 bal.logf("%s unreferenced (have>want=0, new)", bal.stats.unref)
929 bal.logf("%s garbage (have>want=0, old)", bal.stats.garbage)
930 for _, class := range bal.classes {
931 cs := bal.stats.classStats[class]
933 bal.logf("storage class %q: %s desired", class, cs.desired)
934 bal.logf("storage class %q: %s short", class, cs.short)
935 bal.logf("storage class %q: %s surplus", class, cs.surplus)
936 bal.logf("storage class %q: %s unachievable", class, cs.unachievable)
939 bal.logf("%s total commitment (excluding unreferenced)", bal.stats.desired)
940 bal.logf("%s total usage", bal.stats.current)
942 for _, srv := range bal.KeepServices {
943 bal.logf("%s: %v\n", srv, srv.ChangeSet)
946 bal.printHistogram(60)
950 func (bal *Balancer) printHistogram(hashColumns int) {
951 bal.logf("Replication level distribution (counting N replicas on a single server as N):")
953 for _, count := range bal.stats.replHistogram {
954 if maxCount < count {
958 hashes := strings.Repeat("#", hashColumns)
959 countWidth := 1 + int(math.Log10(float64(maxCount+1)))
960 scaleCount := 10 * float64(hashColumns) / math.Floor(1+10*math.Log10(float64(maxCount+1)))
961 for repl, count := range bal.stats.replHistogram {
962 nHashes := int(scaleCount * math.Log10(float64(count+1)))
963 bal.logf("%2d: %*d %s", repl, countWidth, count, hashes[:nHashes])
967 // CheckSanityLate checks for configuration and runtime errors after
968 // GetCurrentState() and ComputeChangeSets() have finished.
970 // If it returns an error, it is dangerous to run any Commit methods.
971 func (bal *Balancer) CheckSanityLate() error {
972 if bal.errors != nil {
973 for _, err := range bal.errors {
974 bal.logf("deferred error: %v", err)
976 return fmt.Errorf("cannot proceed safely after deferred errors")
979 if bal.collScanned == 0 {
980 return fmt.Errorf("received zero collections")
984 bal.BlockStateMap.Apply(func(_ arvados.SizedDigest, blk *BlockState) {
985 for _, desired := range blk.Desired {
993 return fmt.Errorf("zero blocks have desired replication>0")
996 if dr := bal.DefaultReplication; dr < 1 {
997 return fmt.Errorf("Default replication (%d) is less than 1", dr)
1000 // TODO: no two services have identical indexes
1001 // TODO: no collisions (same md5, different size)
1005 // CommitPulls sends the computed lists of pull requests to the
1006 // keepstore servers. This has the effect of increasing replication of
1007 // existing blocks that are either underreplicated or poorly
1008 // distributed according to rendezvous hashing.
1009 func (bal *Balancer) CommitPulls(c *arvados.Client) error {
1010 defer bal.time("send_pull_lists", "wall clock time to send pull lists")()
1011 return bal.commitAsync(c, "send pull list",
1012 func(srv *KeepService) error {
1013 return srv.CommitPulls(c)
1017 // CommitTrash sends the computed lists of trash requests to the
1018 // keepstore servers. This has the effect of deleting blocks that are
1019 // overreplicated or unreferenced.
1020 func (bal *Balancer) CommitTrash(c *arvados.Client) error {
1021 defer bal.time("send_trash_lists", "wall clock time to send trash lists")()
1022 return bal.commitAsync(c, "send trash list",
1023 func(srv *KeepService) error {
1024 return srv.CommitTrash(c)
1028 func (bal *Balancer) commitAsync(c *arvados.Client, label string, f func(srv *KeepService) error) error {
1029 errs := make(chan error)
1030 for _, srv := range bal.KeepServices {
1031 go func(srv *KeepService) {
1033 defer func() { errs <- err }()
1034 label := fmt.Sprintf("%s: %v", srv, label)
1037 err = fmt.Errorf("%s: %v", label, err)
1042 for range bal.KeepServices {
1043 if err := <-errs; err != nil {
1052 func (bal *Balancer) logf(f string, args ...interface{}) {
1053 if bal.Logger != nil {
1054 bal.Logger.Printf(f, args...)
1058 func (bal *Balancer) time(name, help string) func() {
1059 observer := bal.Metrics.DurationObserver(name+"_seconds", help)
1061 bal.Logger.Printf("%s: start", name)
1063 dur := time.Since(t0)
1064 observer.Observe(dur.Seconds())
1065 bal.Logger.Printf("%s: took %vs", name, dur.Seconds())
1069 // Rendezvous hash sort function. Less efficient than sorting on
1070 // precomputed rendezvous hashes, but also rarely used.
1071 func rendezvousLess(i, j string, blkid arvados.SizedDigest) bool {
1072 a := md5.Sum([]byte(string(blkid[:32]) + i))
1073 b := md5.Sum([]byte(string(blkid[:32]) + j))
1074 return bytes.Compare(a[:], b[:]) < 0