Skip to content

Conversation

@swamirishi
Copy link
Contributor

What changes were proposed in this pull request?

Currently all garbage collection threads peek into previous snapshots to check whether a particular deleted key/deleted Directory/ RenameEntry can be removed from the system. All of these operations happen without taking locks because of which all of these threads could be working on an inconsistent snapshot data.

This patch implements a general utility class which would be responsible for opening last N snapshots from the snapshot chain corresponding to the keys bucket where N is going to be defined by the garbage collection thread based on whether if it is reclaiming a deletedKey/deletedDirectory/RenameEntry after taking a read lock on all of these snapshots. If there are lesser than N previous snapshot in the chain, only the existing snapshots would be considered and rest of them would be just marked as null. This utility class should fail opening the snapshot if any of these snaphshots are not active, since a deleted snapshot could be in an inconsistent state.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-12560

How was this patch tested?

Adding unit tests

…napshots

Change-Id: If1a980e5550dda6775c3af53b14a1eab71206a43
Change-Id: Ia8cd0c367e54c978d7534e7bd3702803c8cdf38e
Change-Id: I772783d23bf250ca6d468bf6326a09a49c614eb9
Change-Id: I1c1ad2f536ead10a92caa7eec0b8d891289a0257
Change-Id: Ib70ac6c3feade6dc3c24b34c537aa551bbf31062
@ivandika3 ivandika3 added the snapshot https://issues.apache.org/jira/browse/HDDS-6517 label Mar 21, 2025
Copy link
Contributor

@hemantk-12 hemantk-12 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @swamirishi for the improvement and single reclaimable filter.

I mainly reviewed bd5b0c6 because it is on top of #8052

package org.apache.hadoop.ozone.util;

/**
*
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: please remove this unnecessary line.

Comment on lines 73 to 76
* @param currentSnapshotInfo : If null the deleted keys in AOS needs to be processed, hence the latest snapshot
* in the snapshot chain corresponding to bucket key needs to be processed.
* @param keyManager : KeyManager corresponding to snapshot or AOS.
* @param lock : Lock for Active OM.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please add all the parameters or remove all of them from the JavaDoc comment.

}

// Reversing list to get the correct order in chain. To ensure locking order is as per the chain ordering.
Collections.reverse(snapshotInfos);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yeah we don't need this anymore

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We rely on this order to be right right for the garbage collection algorithm to work correctly. We are keeping the snapshot chain in the sorted order. If you want me to change that I can do that as well.

Comment on lines 166 to 167
throw new IOException("Lock acquisition failed for last N snapshots : " +
expectedLastNSnapshotsInChain + " " + currentSnapshotInfo);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
throw new IOException("Lock acquisition failed for last N snapshots : " +
expectedLastNSnapshotsInChain + " " + currentSnapshotInfo);
throw new IOException("Lock acquisition failed for last N snapshots: " +
expectedLastNSnapshotsInChain + ", " + currentSnapshotInfo);

qq: Is adding expectedLastNSnapshotsInChain to the log message useful?

Change-Id: I5b4eedefd85881933572df1a6c8b2960a2cc9278
Change-Id: I6a5e9ab92c10164288c090a8b748383a9430ee69
Change-Id: I411b2d0a6f3fa0551ec2f306317440f8ed7387c0
Change-Id: Ie7575e9b7ccb1eb9c2b6d875e9d88e4005d12026
Change-Id: I53fa063a7268e23870b3a7a6851e9920b81ed241
Change-Id: I298fc6d96335b5cb9c79f46901f0d89dc2829362
* @param numberOfPreviousSnapshotsFromChain : number of previous snapshots to be initialized.
*/
public ReclaimableFilter(OzoneManager ozoneManager, OmSnapshotManager omSnapshotManager,
SnapshotChainManager snapshotChainManager,
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Move SnapshotChainManager inside OmSnapshotManager

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I will create a follow up task item. This could be a good newbie jira

Change-Id: I9c80b73b6b72d7214edb3e92ae32799a4308f223
Change-Id: I20a877c63a968d9bf912e12cfd10c7aebc7a25bb
Change-Id: I7f5ab7eccbade1f71515e86c9667ab9c1948189e
Change-Id: I62416787dbc587f393f795766d555ea536b8cda5

# Conflicts:
#	hadoop-ozone/ozone-manager/src/test/java/org/apache/hadoop/ozone/om/snapshot/TestMultiSnapshotLocks.java
Copy link
Contributor

@adoroszlai adoroszlai left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @swamirishi for the patch. Only some nits on generic items, haven't reviewed core logic.

* <R> – the type of the result of the function
* <E> - the type of exception thrown.
*/
public interface CheckedFunction<T, R, E extends Exception> {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please use org.apache.ratis.util.function.CheckedFunction instead

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

+1 .We can reuse the existing interface.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

done

? SnapshotUtils.getLatestSnapshotInfo(volume, bucket, ozoneManager, snapshotChainManager)
: SnapshotUtils.getPreviousSnapshot(ozoneManager, snapshotChainManager, currentSnapshotInfo);
SnapshotInfo snapshotInfo = expectedPreviousSnapshotInfo;
while (tmpValidationSnapshotInfos.size() < numberOfPreviousSnapshotsFromChain) {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If snapshotInfo is null there is no previous snapshot, Can we terminate this loop early?

Suggested change
while (tmpValidationSnapshotInfos.size() < numberOfPreviousSnapshotsFromChain) {
while (snapshotInfo != null && tmpValidationSnapshotInfos.size() < numberOfPreviousSnapshotsFromChain) {

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Even if the snapshot info is null. We still want the previous SnapshotInfos list to be equal to the number of numberOfPreviousSnapshotsFromChain. When the snapshotInfo is null, we want the list to be filled with the required null values.

swamirishi and others added 2 commits April 17, 2025 08:23
Change-Id: I4e57e43596a9d90ecb2dad62883b746d8cdf992e
…ne/om/snapshot/filter/TestAbstractReclaimableFilter.java

Co-authored-by: Doroszlai, Attila <[email protected]>
swamirishi and others added 7 commits April 17, 2025 08:29
Change-Id: I7114755259f35362cc7a5db527663e39e38c8b7a
Change-Id: Ife615fc19c66ec5db1558d8ea930cc45f65f41b6
Change-Id: If5159bfd6cb61e3c8ab00e8322de290d27d41940
Change-Id: I18a97d75efb94cb2aec91fbdb1aef57ff11fb1a7
…ne/om/snapshot/filter/ReclaimableFilter.java

Co-authored-by: Doroszlai, Attila <[email protected]>
Change-Id: I864704aa4ef0140eb2d3005795cfeb89c721246d
Change-Id: If60fe12bf3a5bc111c62bc86889027fe291c8baa
Copy link
Contributor

@hemantk-12 hemantk-12 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sorry, I forgot to hit the publish button.

* latest snapshot in the snapshot chain corresponding to bucket key needs to be
* processed.
* @param keyManager : KeyManager corresponding to snapshot or Active Metadata Manager.
* @param lock : Lock for Active OM.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it lock for active OM? or more like lock manager?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You are right it is the lock manager for OM

Comment on lines 75 to 77
* @param currentSnapshotInfo : If null the deleted keys in Active Metadata manager needs to be processed, hence the
* latest snapshot in the snapshot chain corresponding to bucket key needs to be
* processed.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why does the second part of this comment mean?

Copy link
Contributor Author

@swamirishi swamirishi Apr 30, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reworded it. lmk if it is better worded now.

Copy link
Contributor

@hemantk-12 hemantk-12 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall changes look good to me. Left some minor comments, otherwise +1.

Change-Id: I443fd757524cfcd35780f06631f06112e33b0f04
Change-Id: Ic1c1920826f5462cb38d36075cffa498814c96c7
Copy link
Contributor

@hemantk-12 hemantk-12 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks @swamirishi for taking care of snapshot garbage collection.

@swamirishi swamirishi merged commit ee32fa5 into apache:master Apr 30, 2025
42 checks passed
@swamirishi
Copy link
Contributor Author

Thanks @hemantk-12 @adoroszlai @aswinshakil for reviewing the patch

vtutrinov pushed a commit to vtutrinov/ozone that referenced this pull request May 13, 2025
aswinshakil added a commit to aswinshakil/ozone that referenced this pull request May 16, 2025
…239-container-reconciliation

Commits: 113 commits

8ceb5c3 Revert "HDDS-11232. Spare InfoBucket RPC call for the FileSystem#getFileStatus calls for the general case. (apache#6988)" (apache#8358)
8ecf4a8 HDDS-12993. Use DatanodeID in ReconNodeManager. (apache#8456)
3e09a11 HDDS-13024. HTTP connect to 0.0.0.0 failed (apache#8439)
c3cb14f HDDS-12914. Bump awssdk to 2.31.40, test ResumableFileDownload (apache#8455)
c6b47e9 HDDS-4677. Document Ozone Ports and Connection End Points (apache#8226)
ae0d757 HDDS-13044. Remove DatanodeDetails#getUuid usages from hdds-common/client/container-service (apache#8462)
2bf3f16 HDDS-12568. Implement MiniOzoneCluster.Service for Recon (apache#8452)
e59d251 HDDS-13046. Add .vscode to .gitignore (apache#8461)
50e6d61 HDDS-13022. Split up exclusive size tracking for key and directory cleanup in SnapshotInfo (apache#8433)
6db647d HDDS-12966. DBDefinitionFactory should not throw InvalidArnException (apache#8454)
fe2875c HDDS-12948. [Snapshot] Increase `ozone.om.fs.snapshot.max.limit` default value to 10k (apache#8376)
215bdc2 HDDS-11904. Fix HealthyPipelineSafeModeRule logic. (apache#8386)
dba7557 HDDS-12989. Throw CodecException for the Codec byte[] methods (apache#8444)
d9c73f0 HDDS-11298. Support owner field in the S3 listBuckets request (apache#8315)
7a5129d HDDS-12810. Check and reserve space atomically in VolumeChoosingPolicy (apache#8360)
d51ccf7 HDDS-13016. Add a getAllNodeCount() method to NodeManager. (apache#8445)
8b27fb9 HDDS-12299. Merge OzoneAclConfig into OmConfig (apache#8383)
96f4f79 HDDS-12501. Remove OMKeyInfo from SST files in backup directory. (apache#8214)
8adb500 HDDS-13021. Make callId unique for each request in AbstractKeyDeletingService (apache#8432)
faeadd6 HDDS-13000. [Snapshot] listSnapshotDiffJobs throws NPE. (apache#8418)
7c957cc HDDS-11954. List keys command gets wrong info from BasicOmKeyInfo (apache#8401)
0697383 HDDS-13027. Workaround for GitHub connection failures (apache#8443)
0d5f933 HDDS-12919. Remove redundant FileLock from ChunkWrite (apache#8435)
395892a HDDS-13012. [Snapshot] Add audit logs for listSnapshotDiffJobs, snapshotDiff and cancelSnapshotDiff. (apache#8425)
d92f0ed HDDS-12653. Add option in `ozone debug log container list` to filter by Health State (apache#8415)
78f8a02 HDDS-12645. Improve output of `ozone debug replicas chunk-info` (apache#8146)
f32470c HDDS-12995. Split ListOptions for better reuse (apache#8413)
cdd1b13 HDDS-13019. Bump jline to 3.30.0 (apache#8428)
1651408 HDDS-13018. Bump common-custom-user-data-maven-extension to 2.0.2 (apache#8429)
8f6222e HDDS-12949. Cleanup SCMSafeModeManager (apache#8390)
adb385f HDDS-13002. Use DatanodeID in ContainerBalancerTask (apache#8421)
2b2a796 HDDS-12969. Use DatanodeID in XceiverClientGrpc (apache#8430)
3c0bafa HDDS-13001. Use DatanodeID in SCMBlockDeletingService. (apache#8420)
8448bc2 HDDS-12187. Use hadoop-dependency-client as parent, not dependency (apache#8416)
582fd03 HDDS-13015. Bump zstd-jni to 1.5.7-3 (apache#8426)
ec12f98 HDDS-12970. Use DatanodeID in Pipeline (apache#8414)
3f48a18 HDDS-12179. Improve Container Log to also capture the Volume Information (apache#7858)
034b497 HDDS-12964. Add minimum required version of Docker Compose to RunningViaDocker page (apache#8385)
c0a6b42 HDDS-12947. Add CodecException (apache#8411)
70949c5 HDDS-11603. Reclaimable Key Filter for Snapshots garbage reclaimation (apache#8392)
08283f3 HDDS-12087. TransactionToDNCommitMap too large causes GC to pause for a long time (apache#8347)
f47df78 HDDS-12958. [Snapshot] Add ACL check regression tests for snapshot operations. (apache#8419)
1cc8445 HDDS-12207. Unify output of ozone debug replicas verify checks (apache#8248)
f087d0b HDDS-12994. Use DatanodeID in ReconSCMDBDefinition. (apache#8417)
d6a7723 HDDS-12776. ozone debug CLI command to list all Duplicate open containers (apache#8409)
c78aeb0 HDDS-12951. EC: Log when falling back to reconstruction read (apache#8408)
412f22d HDDS-12959. Eliminate hdds-hadoop-dependency-server (apache#8384)
df701dc HDDS-12996. Workaround for Docker Compose concurrent map writes (apache#8412)
e6daae4 HDDS-12972. Use DatanodeID in ContainerReplica. (apache#8396)
c1103ae HDDS-12877. Support StorageClass field in the S3 HeadObject request (apache#8351)
4135384 HDDS-12973. Add javadoc for CompactionNode() and make getCompactionNodeGraph return ConcurrentMap (apache#8395)
4f2e13c HDDS-12954. Do not throw IOException for checksum. (apache#8387)
49b8fbd HDDS-12971. Use DatanodeID in Node2PipelineMap (apache#8403)
d2da18f HDDS-12346. Reduce code duplication among TestNSSummaryTask classes (apache#8287)
82b73e3 HDDS-11856. Set DN state machine thread priority higher than command handler thread. (apache#8253)
d4f2734 HDDS-12689. Import BOM for AWS SDK, declare dependencies (apache#8406)
4775e76 HDDS-12975. Fix percentage of blocks deleted in grafana dashboard (apache#8398)
ac0d696 HDDS-12968. [Recon] Fix column visibility issue in Derby during schema upgrade finalization. (apache#8393)
e71dcf6 HDDS-11981. Add annotation for registering feature validator based on a generic version (apache#7603)
254297c HDDS-12562. Reclaimable Directory entry filter for reclaiming deleted directory entries (apache#8055)
4f467c8 HDDS-12978. Remove TestMultipartObjectGet (apache#8400)
a99f207 HDDS-12967. Skip CommonChunkManagerTestCases.testFinishWrite if fuser cannot be started (apache#8389)
d29d76b HDDS-12697. Ozone debug CLI to display details of a single container (apache#8264)
1d1bc88 HDDS-12974. Docker could not parse extra host IP (apache#8397)
7e675d7 HDDS-12053. Make print-log-dag command run locally and offline (apache#8016)
d3faab3 HDDS-12561. Reclaimable Rename entry filter for reclaiming renaming entries (apache#8054)
af1f98c HDDS-10822. Tool to omit raft log in OM. (apache#8154)
3201ca4 HDDS-12952. Make OmSnapshotManager#snapshotLimitCheck thread-safe and consistent (apache#8381)
522c88d HDDS-12963. Clean up io.grpc dependencies (apache#8382)
fa8bd9d HDDS-12916. Support ETag in listObjects response (apache#8356)
fdc77db HDDS-12300. Merge OmUpgradeConfig into OmConfig (apache#8378)
623e144 HDDS-12956. Bump vite to 4.5.14 (apache#8375)
8c8eaf1 HDDS-12944. Reduce timeout for integration check (apache#8374)
40d2e00 HDDS-11141. Avoid log flood due due pipeline close in XceiverServerRatis (apache#8325)
9fe1dba HDDS-12942. Init layout version config should not be public (apache#8373)
452e7aa HDDS-12596. OM fs snapshot max limit is not enforced (apache#8377)
8b095d5 HDDS-12795. Rename heartbeat and first election configuration name (apache#8249)
bee8164 HDDS-12920. Configure log4j to gzip rolled over service log files (apache#8357)
e16a50f HDDS-12934. Split submodule for Freon. (apache#8367)
b1e9511 HDDS-12925. Update datanode volume used space on container deletion (apache#8364)
440bc82 Revert "HDDS-12596. OM fs snapshot max limit is not enforced (apache#8157)"
f345492 HDDS-12596. OM fs snapshot max limit is not enforced (apache#8157)
ee7b1dc HDDS-12901. Introduce EventExecutorMetrics instead of setting the metrics props unsafely (apache#8371)
810e148 HDDS-12939. Remove UnknownPipelineStateException. (apache#8372)
560fcdf HDDS-12728. Add Ozone 2.0.0 to compatibility acceptance tests (apache#8361)
5815a47 HDDS-12933. Remove the table names declared in OmMetadataManagerImpl (apache#8370)
ee32fa5 HDDS-12560. Reclaimable Filter for Snaphost Garbage Collections (apache#8053)
45374ea HDDS-12932. Rewrite OMDBDefinition (apache#8362)
5cb6dd8 HDDS-12575. Set default JUnit5 timeout via property (apache#8348)
8efc0cd HDDS-11633. Delete message body too large, causing SCM to fail writing raft log (apache#8354)
ac9d9fd HDDS-12915. Intermittent failure in testCreatePipelineThrowErrorWithDataNodeLimit (apache#8359)
86039e8 HDDS-12848. Create new submodule for ozone admin (apache#8292)
2d0f8cb HDDS-12833. Remove the CodecRegistry field from DBStoreBuilder (apache#8327)
c71b393 HDDS-12921. UnusedPrivateField violations in tests (apache#8353)
9f3dd01 HDDS-12917. cp: option '--update' doesn't allow an argument (apache#8346)
a14b395 HDDS-12922. Use OMDBDefinition in GeneratorOm and FSORepairTool (apache#8355)
c8a98d6 HDDS-12892. OM Tagging Request incorrectly sets full path as key name for FSO (apache#8345)
ade69e3 HDDS-12649. Include name of volume or bucket in length validation error (apache#8322)
c68308d HDDS-12599. Create an ozone debug CLI command to list all the containers based on final state (apache#8282)
319d5a4 HDDS-12773. bad substitution in bats test (apache#8290)
6f5e02a HDDS-12900. (addendum: fix pmd) Use OMDBDefinition in OmMetadataManagerImpl (apache#8337)
2a7000d HDDS-12900. Use OMDBDefinition in OmMetadataManagerImpl (apache#8337)
9c0c66c HDDS-12915. Mark testCreatePipelineThrowErrorWithDataNodeLimit as flaky
a73e052 HDDS-12907. Enable FieldDeclarationsShouldBeAtStartOfClass PMD rule (apache#8344)
d083f82 HDDS-12905. Move field declarations to start of class in ozone-common (apache#8342)
3f90e1c HDDS-12906. Move field declarations to start of class in ozone-manager module (apache#8343)
403fb97 HDDS-12878. Move field declarations to start of class in tests (apache#8308)
63d5c73 HDDS-12912. Remove deprecated `PipelineManager#closePipeline(Pipeline, boolean)` (apache#8340)
cf1fb88 HDDS-12902. Shutdown executor in CloseContainerCommandHandler and ECReconstructionCoordinator (apache#8341)
825ba02 HDDS-9585. Improve import/export log in ContainerLogger (apache#8330)
b70d35a HDDS-12889. Enable AppendCharacterWithChar PMD rule (apache#8324)
cd308ea HDDS-12904. Move field declarations to start of class in other hdds modules (apache#8336)
4905286 HDDS-12899. Move field declarations to start of class in hdds-server-scm (apache#8332)

Conflicts:
	hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/common/utils/ContainerLogger.java
	hadoop-hdds/container-service/src/test/java/org/apache/hadoop/ozone/container/keyvalue/TestKeyValueHandler.java
	hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/container/ContainerReplica.java
	hadoop-ozone/cli-admin/src/main/java/org/apache/hadoop/hdds/scm/cli/container/ReconcileSubcommand.java
	hadoop-ozone/recon-codegen/src/main/java/org/apache/ozone/recon/schema/ContainerSchemaDefinition.java
swamirishi added a commit to swamirishi/ozone that referenced this pull request Dec 3, 2025
…ctions (apache#8053)

Change-Id: I389410bae55007d8d6d8a64bfbd319509146a0bd
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

snapshot https://issues.apache.org/jira/browse/HDDS-6517

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants