[HUDI-2143] Tweak the default compaction target IO to 500GB when flin… #3238

danny0405 · 2021-07-08T03:55:53Z

…k async compaction is off

Tips

Thank you very much for contributing to Apache Hudi.
Please review https://hudi.apache.org/contributing.html before opening a pull request.

What is the purpose of the pull request

(For example: This pull request adds quick-start document.)

Brief change log

(for example:)

Modify AnnotationLocation checkstyle rule in checkstyle.xml

Verify this pull request

(Please pick either of the following options)

This pull request is a trivial rework / code cleanup without any test coverage.

(or)

This pull request is already covered by existing tests, such as (please describe tests).

(or)

This change added tests and can be verified as follows:

(example:)

Added integration tests for end-to-end.
Added HoodieClientWriteTest to verify the change.
Manually verified the change by running a job locally.

Committer checklist

Has a corresponding JIRA in PR title & commit
Commit message is descriptive of the change
CI is green
Necessary doc changes done or have another open PR
For large changes, please consider breaking it into sub-tasks under an umbrella JIRA.

…k async compaction is off

hudi-bot · 2021-07-08T03:58:07Z

CI report:

51d3c5f Azure: FAILURE

Bot commands

@hudi-bot supports the following commands:

@hudi-bot run travis re-run the last Travis build
@hudi-bot run azure re-run the last Azure build

codecov-commenter · 2021-07-08T06:18:14Z

Codecov Report

Merging #3238 (51d3c5f) into master (16e90d3) will increase coverage by 0.00%.
The diff coverage is 20.00%.

@@            Coverage Diff            @@
##             master    #3238   +/-   ##
=========================================
  Coverage     47.62%   47.63%           
- Complexity     5502     5508    +6     
=========================================
  Files           930      930           
  Lines         41268    41312   +44     
  Branches       4137     4142    +5     
=========================================
+ Hits          19655    19680   +25     
- Misses        19865    19883   +18     
- Partials       1748     1749    +1

Flag	Coverage Δ
hudicli	`39.97% <ø> (ø)`
hudiclient	`34.58% <ø> (ø)`
hudicommon	`48.59% <ø> (ø)`
hudiflink	`59.53% <20.00%> (-0.06%)`	⬇️
hudihadoopmr	`51.29% <ø> (ø)`
hudisparkdatasource	`67.24% <ø> (-0.09%)`	⬇️
hudisync	`54.48% <ø> (ø)`
huditimelineservice	`64.07% <ø> (ø)`
hudiutilities	`58.60% <ø> (+0.03%)`	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...java/org/apache/hudi/table/HoodieTableFactory.java	`86.84% <20.00%> (-4.83%)`	⬇️
...pache/hudi/internal/HoodieWriterCommitMessage.java	`100.00% <0.00%> (ø)`
...spark3/internal/HoodieDataSourceInternalTable.java	`0.00% <0.00%> (ø)`
...nal/HoodieDataSourceInternalBatchWriteBuilder.java	`0.00% <0.00%> (ø)`
...apache/hudi/utilities/deltastreamer/DeltaSync.java	`71.47% <0.00%> (+0.33%)`	⬆️
.../hudi/internal/HoodieDataSourceInternalWriter.java	`80.00% <0.00%> (+3.80%)`	⬆️
...3/internal/HoodieDataSourceInternalBatchWrite.java	`83.87% <0.00%> (+8.87%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 16e90d3...51d3c5f. Read the comment docs.

garyli1019 · 2021-07-08T09:22:35Z

hudi-flink/src/main/java/org/apache/hudi/table/HoodieTableFactory.java

      conf.setInteger(FlinkOptions.ARCHIVE_MIN_COMMITS, commitsToRetain + 10);
      conf.setInteger(FlinkOptions.ARCHIVE_MAX_COMMITS, commitsToRetain + 20);
    }
+    if (conf.getBoolean(FlinkOptions.COMPACTION_SCHEDULE_ENABLED)


why not just respect the config user-defined? this change the config without noticing anyone

I have decided that if the option value is as the default.

IMO, @garyli1019 's opinion is reasonable. @danny0405 I know what's your purpose. Can we define an explicit config option for this? So that users can know this detailed information.

We have respected the user specified option, the value is rewritten only when the default value is set.

Yes, but the user may not know it would be rewritten under some condition like this?

Why the user needs to know that ? They can set up the target IO by themself if they want BTW.

The rewritten value is a constant value. Why not provide an SYNC_COMPACTION_TARGET_IO option and make it to be 500GB?

There is already one, the default value is 5GB for online compaction, this patch fix the offline compaction with default value as 500GB.

sorry, a bit late on this. I don't have a strong objection here but slightly incline to not turning config inside the code.
My point is that tweaking config is ok if the user wrote their own connector or wanted to reduce the overhead of having too many configs in their app. I do this in my app as well. But this is the official connector of Hudi, so I think we should keep the config as how it is.

danny0405 · 2021-07-08T09:38:16Z

hudi-flink/src/main/java/org/apache/hudi/table/HoodieTableFactory.java

    }
+    if (conf.getBoolean(FlinkOptions.COMPACTION_SCHEDULE_ENABLED)
+        && !conf.getBoolean(FlinkOptions.COMPACTION_ASYNC_ENABLED)
+        && FlinkOptions.isDefaultValueDefined(conf, FlinkOptions.COMPACTION_TARGET_IO)) {


See this line.

yanghua

LGTM

change the insret overwrte return type [HUDI-1860] Test wrapper for insert_overwrite and insert_overwrite_table [HUDI-2084] Resend the uncommitted write metadata when start up (apache#3168) Co-authored-by: 喻兆靖 <[email protected]> [HUDI-2081] Move schema util tests out from TestHiveSyncTool (apache#3166) [HUDI-2094] Supports hive style partitioning for flink writer (apache#3178) [HUDI-2097] Fix Flink unable to read commit metadata error (apache#3180) [HUDI-2085] Support specify compaction paralleism and compaction target io for flink batch compaction (apache#3169) [HUDI-2092] Fix NPE caused by FlinkStreamerConfig#writePartitionUrlEncode null value (apache#3176) [HUDI-2006] Adding more yaml templates to test suite (apache#3073) [HUDI-2103] Add rebalance before index bootstrap (apache#3185) Co-authored-by: 喻兆靖 <[email protected]> [HUDI-1944] Support Hudi to read from committed offset (apache#3175) * [HUDI-1944] Support Hudi to read from committed offset * [HUDI-1944] Adding group option to KafkaResetOffsetStrategies * [HUDI-1944] Update Exception msg [HUDI-2052] Support load logFile in BootstrapFunction (apache#3134) Co-authored-by: 喻兆靖 <[email protected]> [HUDI-89] Add configOption & refactor all configs based on that (apache#2833) Co-authored-by: Wenning Ding <[email protected]> [MINOR] Update .asf.yaml to codify notification settings, turn on jira comments, gh discussions (apache#3164) - Turn on comment for jira, so we can track PR activity better - Create a notification settings that match https://gitbox.apache.org/schemes.cgi?hudi - Try and turn on "discussions" on Github, to experiment [MINOR] Fix broken build due to FlinkOptions (apache#3198) [HUDI-2088] Missing Partition Fields And PreCombineField In Hoodie Properties For Table Written By Flink (apache#3171) [MINOR] Add Documentation to KEYGENERATOR_TYPE_PROP (apache#3196) [HUDI-2105] Compaction Failed For MergeInto MOR Table (apache#3190) [HUDI-2051] Enable Hive Sync When Spark Enable Hive Meta For Spark Sql (apache#3126) [HUDI-2112] Support reading pure logs file group for flink batch reader after compaction (apache#3202) [HUDI-2114] Spark Query MOR Table Written By Flink Return Incorrect Timestamp Value (apache#3208) [HUDI-2121] Add operator uid for flink stateful operators (apache#3212) [HUDI-2123] Exception When Merge With Null-Value Field (apache#3214) [HUDI-2124] A Grafana dashboard for HUDI. (apache#3216) [HUDI-2057] CTAS Generate An External Table When Create Managed Table (apache#3146) [HUDI-1930] Bootstrap support configure KeyGenerator by type (apache#3170) * [HUDI-1930] Bootstrap support configure KeyGenerator by type [HUDI-2116] Support batch synchronization of partition datas to hive metastore to avoid oom problem (apache#3209) [HUDI-2126] The coordinator send events to write function when there are no data for the checkpoint (apache#3219) [HUDI-2127] Initialize the maxMemorySizeInBytes in log scanner (apache#3220) [HUDI-2058]support incremental query for insert_overwrite_table/insert_overwrite operation on cow table (apache#3139) [HUDI-2129] StreamerUtil.medianInstantTime should return a valid date time string (apache#3221) [HUDI-2131] Exception Throw Out When MergeInto With Decimal Type Field (apache#3224) [HUDI-2122] Improvement in packaging insert into smallfiles (apache#3213) [HUDI-2132] Make coordinator events as POJO for efficient serialization (apache#3223) [HUDI-2106] Fix flink batch compaction bug while user don't set compaction tasks (apache#3192) [HUDI-2133] Support hive1 metadata sync for flink writer (apache#3225) [HUDI-2089]fix the bug that metatable cannot support non_partition table (apache#3182) [HUDI-2028] Implement RockDbBasedMap as an alternate to DiskBasedMap in ExternalSpillableMap (apache#3194) Co-authored-by: Rajesh Mahindra <[email protected]> [HUDI-2135] Add compaction schedule option for flink (apache#3226) [HUDI-2055] Added deltastreamer metric for time of lastSync (apache#3129) [HUDI-2046] Loaded too many classes like sun/reflect/GeneratedSerializationConstructorAccessor in JVM metaspace (apache#3121) Loaded too many classes when use kryo of spark to hudi Co-authored-by: weiwei.duan <[email protected]> [HUDI-1996] Adding functionality to allow the providing of basic auth creds for confluent cloud schema registry (apache#3097) * adding support for basic auth with confluent cloud schema registry [HUDI-2093] Fix empty avro schema path caused by duplicate parameters (apache#3177) * [HUDI-2093] Fix empty avro schema path caused by duplicate parameters * rename shcmea option key * fix doc * rename var name [HUDI-2113] Fix integration testing failure caused by sql results out of order (apache#3204) [HUDI-2016] Fixed bootstrap of Metadata Table when some actions are in progress. (apache#3083) Metadata Table cannot be bootstrapped when any action is in progress. This is detected by the presence of inflight or requested instants. The bootstrapping is initiated in preWrite and postWrite of each commit. So bootstrapping will be retried again until it succeeds. Also added metrics for when the bootstrapping fails or a table is re-bootstrapped. This will help detect tables which are not getting bootstrapped. [HUDI-2140] Fixed the unit test TestHoodieBackedMetadata.testOnlyValidPartitionsAdded. (apache#3234) [HUDI-2115] FileSlices in the filegroup is not descending by timestamp (apache#3206) [HUDI-1104] Adding support for UserDefinedPartitioners and SortModes to BulkInsert with Rows (apache#3149) [HUDI-2069] Refactored String constants (apache#3172) [HUDI-1105] Adding dedup support for Bulk Insert w/ Rows (apache#2206) [HUDI-2134]Add generics to avoif forced conversion in BaseSparkCommitActionExecutor#partition (apache#3232) [HUDI-2009] Fixing extra commit metadata in row writer path (apache#3075) [HUDI-2099]hive lock which state is WATING should be released, otherwise this hive lock will be locked forever (apache#3186) [MINOR] Fix build broken from apache#3186 (apache#3245) [HUDI-2136] Fix conflict when flink-sql-connector-hive and hudi-flink-bundle are both in flink lib (apache#3227) [HUDI-2087] Support Append only in Flink stream (apache#3174) Co-authored-by: 喻兆靖 <[email protected]> UnitTest for deltaSync Removing cosmetic changes and reuse function for insert_overwrite_table unit test intial unit test for the insert_overwrite and insert_over_write_table Adding failed test code for insert_overwrite Revert "[HUDI-2087] Support Append only in Flink stream (apache#3174)" (apache#3251) This reverts commit 3715267. [HUDI-2147] Remove unused class AvroConvertor in hudi-flink (apache#3243) [MINOR] Fix some wrong assert reasons (apache#3248) [HUDI-2087] Support Append only in Flink stream (apache#3252) Co-authored-by: 喻兆靖 <[email protected]> [HUDI-2143] Tweak the default compaction target IO to 500GB when flink async compaction is off (apache#3238) [HUDI-2142] Support setting bucket assign parallelism for flink write task (apache#3239) [HUDI-1483] Support async clustering for deltastreamer and Spark streaming (apache#3142) - Integrate async clustering service with HoodieDeltaStreamer and HoodieStreamingSink - Added methods in HoodieAsyncService to reuse code [HUDI-2107] Support Read Log Only MOR Table For Spark (apache#3193) [HUDI-2144]Bug-Fix:Offline clustering(HoodieClusteringJob) will cause insert action losing data (apache#3240) * fixed * add testUpsertPartitionerWithSmallFileHandlingAndClusteringPlan ut * fix CheckStyle Co-authored-by: yuezhang <[email protected]> [MINOR] Fix EXTERNAL_RECORD_AND_SCHEMA_TRANSFORMATION config (apache#3250) [HUDI-2168] Fix for AccessControlException for anonymous user (apache#3264) [HUDI-2045] Support Read Hoodie As DataSource Table For Flink And DeltaStreamer test with insert-overwrite and insert-overwrite-table removing hardcoded action to pass the unit test [HUDI-1969] Support reading logs for MOR Hive rt table (apache#3033) [HUDI-2171] Add parallelism conf for bootstrap operator using delta-commit for insert_overwrite

…k async compaction is off (apache#3238)

[HUDI-2143] Tweak the default compaction target IO to 500GB when flin…

51d3c5f

…k async compaction is off

danny0405 closed this Jul 8, 2021

danny0405 reopened this Jul 8, 2021

garyli1019 reviewed Jul 8, 2021

View reviewed changes

danny0405 commented Jul 8, 2021

View reviewed changes

yanghua self-assigned this Jul 10, 2021

yanghua approved these changes Jul 10, 2021

View reviewed changes

yanghua merged commit 942a024 into apache:master Jul 10, 2021

ghost pushed a commit to shivagowda/hudi that referenced this pull request Jul 16, 2021

[HUDI-2143] Tweak the default compaction target IO to 500GB when flin…

2c2d2f6

…k async compaction is off (apache#3238)

ghost pushed a commit to shivagowda/hudi that referenced this pull request Aug 1, 2021

[HUDI-2143] Tweak the default compaction target IO to 500GB when flin…

06581f5

…k async compaction is off (apache#3238)

[HUDI-2143] Tweak the default compaction target IO to 500GB when flin… #3238

[HUDI-2143] Tweak the default compaction target IO to 500GB when flin… #3238

Uh oh!

Conversation

danny0405 commented Jul 8, 2021

Tips

What is the purpose of the pull request

Brief change log

Verify this pull request

Committer checklist

Uh oh!

hudi-bot commented Jul 8, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

CI report:

Uh oh!

codecov-commenter commented Jul 8, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yanghua Jul 9, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yanghua left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

hudi-bot commented Jul 8, 2021 •

edited

Loading

codecov-commenter commented Jul 8, 2021 •

edited

Loading

yanghua Jul 9, 2021 •

edited

Loading