[BREAKING] Perf(restore): Implement map-reduce based restore #7664

ahsanbarkati · 2021-03-27T22:20:35Z

This PR implements map-reduce based online restore. The offline
restore has been removed. It also implements backup-export on
top of map-reduce based restore.

This change is

- Make restore respect the drop operations - Fix type and schema issue

- Implement map-reduce based backup-export - Make restore namespace drop aware - Add function for offline restore tests - All offline restore tests pass

- Remove offline restore cli tool - Fix some tests

- Fix oss build - Fix export

manishrjain

Reviewed 24 of 34 files at r1, 16 of 16 files at r2.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @ahsanbarkati, @pawanrawal, and @vvbalaji-dgraph)

ee/backup/run.go, line 292 at r2 (raw file):

		}
		if err := <-errCh; err != nil {
			errors.Wrap(err, "Failed to process reduced buffers")

return

This commit implements map-reduce based online restore. The offline restore has been removed. It also implements backup-export on top of map-reduce based restore. Co-authored-by: Manish R Jain <[email protected]> (cherry picked from commit 98f6828)

fix(lsbackup): Fix profiler in lsBackup (#7729) Bring back "perf(Backup): Improve backup performance (#7601)" Opt(Backup): Make backups faster (#7680) Fix s3 backup copy (#7669) [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) Perf(restore): Implement map-reduce based restore (#7664) feat(backup): Merge backup refactoring Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes in the case of restore. following commits are cherry-picked (in reverse order): * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes in the case of restore. following commits are cherry-picked (in reverse order): * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)" * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix: fixing graphql schema update when the data is restored + skipping /probe/graphql from audit (#7925) * Don't ban namespace in export_backup

This commit is a major rewrite of backup and online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes. cherry-pick PR #7753 following commits are cherry-picked (in reverse order): * Don't ban namespace in export_backup * fix: fixing graphql schema update when the data is restored + skipping /probe/graphql from audit (#7925) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of backup and online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes. cherry-pick PR #7753 following commits are cherry-picked (in reverse order): * reset the kv.StreamId before sending to stream writer (#7833) (#7837) * Don't ban namespace in export_backup * fix: fixing graphql schema update when the data is restored + skipping /probe/graphql from audit (#7925) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of backup and online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes. cherry-pick PR #7753 following commits are cherry-picked (in reverse order): * fix(backup): Fix full backup request (#7932) (#7933) * fix: fixing graphql schema update when the data is restored + * fix(restore): return nil if there is error (#7899) skipping /probe/graphql from audit (#7925) * Don't ban namespace in export_backup * reset the kv.StreamId before sending to stream writer (#7833) (#7837) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of backup and online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes. cherry-pick PR #7753 following commits are cherry-picked (in reverse order): * opt(restore): Sort the buffer before spinning the writeToDisk goroutine (#7984) (#7996) * fix(backup): Fix full backup request (#7932) (#7933) * fix: fixing graphql schema update when the data is restored + * fix(restore): return nil if there is error (#7899) skipping /probe/graphql from audit (#7925) * Don't ban namespace in export_backup * reset the kv.StreamId before sending to stream writer (#7833) (#7837) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of backup and online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes. cherry-pick PR #7753 following commits are cherry-picked (in reverse order): * opt(restore): Sort the buffer before spinning the writeToDisk goroutine (#7984) (#7996) * fix(backup): Fix full backup request (#7932) (#7933) * fix: fixing graphql schema update when the data is restored + skipping /probe/graphql from audit (#7925) * fix(restore): return nil if there is error (#7899) * Don't ban namespace in export_backup * reset the kv.StreamId before sending to stream writer (#7833) (#7837) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of backup and online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes. cherry-pick PR #7753 following commits are cherry-picked (in reverse order): * opt(restore): Sort the buffer before spinning the writeToDisk goroutine (#7984) (#7996) * fix(backup): Fix full backup request (#7932) (#7933) * fix: fixing graphql schema update when the data is restored + * fix(restore): return nil if there is error (#7899) skipping /probe/graphql from audit (#7925) * Don't ban namespace in export_backup * reset the kv.StreamId before sending to stream writer (#7833) (#7837) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

This commit is a major rewrite of backup and online restore code. It used to use KVLoader in badger. Now it instead uses StreamWriter that is much faster for writes. cherry-pick PR #7753 following commits are cherry-picked (in reverse order): * opt(restore): Sort the buffer before spinning the writeToDisk goroutine (#7984) (#7996) * fix(backup): Fix full backup request (#7932) (#7933) * fix: fixing graphql schema update when the data is restored + skipping /probe/graphql from audit (#7925) * fix(restore): return nil if there is error (#7899) * Don't ban namespace in export_backup * reset the kv.StreamId before sending to stream writer (#7833) (#7837) * fix(restore): Bump uid and namespace after restore (#7790) (#7800) * fix(ee): GetKeys should return an error (#7713) (#7797) * fix(backup): Free the UidPack after use (#7786) * fix(export-backup): Fix double free in export backup (#7780) (#7783) * fix(lsbackup): Fix profiler in lsBackup (#7729) * Bring back "perf(Backup): Improve backup performance (#7601)" * Opt(Backup): Make backups faster (#7680) * Fix s3 backup copy (#7669) * [BREAKING] Opt(Restore): Optimize Restore's new map-reduce based design (#7666) * Perf(restore): Implement map-reduce based restore (#7664) * feat(backup): Merge backup refactoring * Revert "perf(Backup): Improve backup performance (#7601)"

manishrjain and others added 7 commits March 20, 2021 18:03

Consolidated all Manifest stuff into one file. Disabled offline restore.

6af1f24

Fix up mapper marshal

631b3b8

Code for backup reduce as well.

a3fca6b

Fix online restore

d61e0cf

- Make restore respect the drop operations - Fix type and schema issue

Fix restore and implement backup-export

a550a7c

- Implement map-reduce based backup-export - Make restore namespace drop aware - Add function for offline restore tests - All offline restore tests pass

Merge branch 'master' into mrjn/restore

00f42ae

Fix(restore): Remove offline restore cli tool

b5d60df

- Remove offline restore cli tool - Fix some tests

ahsanbarkati requested review from manishrjain, pawanrawal and vvbalaji-dgraph as code owners March 27, 2021 22:20

github-actions bot added area/bulk-loader Issues related to bulk loading. area/enterprise Related to proprietary features area/graphql Issues related to GraphQL support on Dgraph. area/live-loader Issues related to live loading. area/testing Testing related issues labels Mar 27, 2021

Some Minor fixes

4fce69c

- Fix oss build - Fix export

ahsanbarkati force-pushed the mrjn/restore branch from 695cb6c to 4fce69c Compare March 28, 2021 07:58

Enforce restoreTs while reduce

3246325

ahsanbarkati force-pushed the mrjn/restore branch 2 times, most recently from 576aca3 to 3246325 Compare March 28, 2021 11:49

manishrjain approved these changes Mar 29, 2021

View reviewed changes

ahsanbarkati added 2 commits March 29, 2021 21:16

Fix error return

1657abf

Remove stale code

d176edd

ahsanbarkati merged commit 98f6828 into master Mar 29, 2021

ahsanbarkati deleted the mrjn/restore branch March 29, 2021 16:04

ahsanbarkati changed the title ~~Perf(restore): Implement map-reduce based restore~~ [BREAKING] Perf(restore): Implement map-reduce based restore Mar 29, 2021

mangalaman93 mentioned this pull request Dec 13, 2022

fix(backup): use StreamWriter instead of KVLoader during backup restore #8510

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[BREAKING] Perf(restore): Implement map-reduce based restore #7664

[BREAKING] Perf(restore): Implement map-reduce based restore #7664

ahsanbarkati commented Mar 27, 2021 •

edited by manishrjain

Loading

manishrjain left a comment

[BREAKING] Perf(restore): Implement map-reduce based restore #7664

[BREAKING] Perf(restore): Implement map-reduce based restore #7664

Conversation

ahsanbarkati commented Mar 27, 2021 • edited by manishrjain Loading

manishrjain left a comment

Choose a reason for hiding this comment

ahsanbarkati commented Mar 27, 2021 •

edited by manishrjain

Loading