pre-configuration based assignment by jasperjiaguo · Pull Request #11578 · apache/pinot

jasperjiaguo · 2023-09-12T22:12:19Z

https://docs.google.com/document/d/1xxPkGPxyY21gAkFi9gtFDeSzEXjPjp-IQW70kHynsL8/edit#heading=h.nit8xkm24pdw

Assign instances based on a pre-configured instance assignment map. The assignment will always respect the mirrored servers in the pre-configured map. E.g.

  uplift from 15 instances in 3 replicas to 21 instance in 3 replicas
  21 instances in 4 pools
  Pre-configured partitioning:
          RG1    RG2    RG3
    Host  0      1      2
    Host  3      4      5
    Host  6      7      8
    Host  9      10     11
    Host  12     13     14
    Host  15     16     17
    Host  18     19     20
  
  Existing configured partitioning:
          RG1    RG2    RG3
    Host  0      6      2
    Host  12     7      14
    Host  1      4      5
    Host  3      13     8
    Host  9      10     11
  
  Final assignment for this table:
          RG1    RG2    RG3
    Host  0      1      2
    Host  12     13     14
    Host  3      4      5
    Host  6      7      8
    Host  9      10     11
    Host  15     16     17
    Host  18     19     20

In the above assignment, the mirrored servers (row) in the pre-configured partitioning will always be respected when converging the existing assignment to the final assignment, meanwhile, we preserve the existing assignment as much as we can in this process.

Tested using randomized testing scheme with 10000000 iterations. No errors.

codecov-commenter · 2023-09-12T23:11:01Z

Codecov Report

Attention: Patch coverage is 67.13781% with 93 lines in your changes missing coverage. Please review.

Project coverage is 61.45%. Comparing base (e826a23) to head (8479b70).
Report is 2564 commits behind head on master.

Files with missing lines	Patch %	Lines
...ller/api/resources/PinotTenantRestletResource.java	0.00%	25 Missing ⚠️
...ntroller/helix/core/rebalance/TableRebalancer.java	25.00%	20 Missing and 1 partial ⚠️
...ources/PinotInstanceAssignmentRestletResource.java	22.72%	16 Missing and 1 partial ⚠️
...ance/MirrorServerSetInstancePartitionSelector.java	92.94%	1 Missing and 11 partials ⚠️
...ntroller/helix/core/PinotHelixResourceManager.java	38.46%	7 Missing and 1 partial ⚠️
...he/pinot/segment/local/utils/TableConfigUtils.java	50.00%	3 Missing and 3 partials ⚠️
...mmon/assignment/InstanceAssignmentConfigUtils.java	0.00%	3 Missing ⚠️
...ent/instance/InstancePartitionSelectorFactory.java	50.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##             master   #11578      +/-   ##
============================================
- Coverage     63.20%   61.45%   -1.75%     
- Complexity     1141     1147       +6     
============================================
  Files          2343     2376      +33     
  Lines        126343   128744    +2401     
  Branches      19439    19899     +460     
============================================
- Hits          79854    79122     -732     
- Misses        40821    43898    +3077     
- Partials       5668     5724      +56

Flag	Coverage Δ
custom-integration1	`<0.01% <0.00%> (-0.01%)`	⬇️
integration	`<0.01% <0.00%> (-0.01%)`	⬇️
integration1	`<0.01% <0.00%> (-0.01%)`	⬇️
integration2	`0.00% <0.00%> (ø)`
java-11	`61.41% <67.13%> (-1.76%)`	⬇️
java-17	`?`
java-20	`?`
java-21	`61.30% <67.13%> (?)`
skip-bytebuffers-false	`61.44% <67.13%> (?)`
skip-bytebuffers-true	`61.28% <67.13%> (?)`
temurin	`61.45% <67.13%> (-1.75%)`	⬇️
unittests	`61.45% <67.13%> (-1.75%)`	⬇️
unittests1	`46.62% <11.76%> (-20.77%)`	⬇️
unittests2	`27.68% <66.43%> (+13.28%)`	⬆️

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

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

Jackie-Jiang · 2023-09-17T20:23:08Z

What is the difference between this and #8989?

add tenant partition map rest resource fix logic refactor and fix some logic pre-configuration based assignment

Jackie-Jiang · 2023-10-24T18:45:06Z

Do you have a design doc for this? Trying to understand what we want to achieve in this PR

vvivekiyer

Flushing out partial comments. I'll review the rest of the files meanwhile.

...spi/src/main/java/org/apache/pinot/spi/config/table/assignment/InstanceAssignmentConfig.java

pinot-core/src/main/java/org/apache/pinot/core/auth/Actions.java

...ller/src/main/java/org/apache/pinot/controller/api/resources/PinotTenantRestletResource.java

...n/java/org/apache/pinot/controller/api/resources/PinotInstanceAssignmentRestletResource.java

siddharthteotia · 2023-10-24T20:56:00Z

Do you have a design doc for this? Trying to understand what we want to achieve in this PR

Let's meet to share our context on where we are coming from and the overall goal here. We will setup a meeting

vvivekiyer · 2023-10-25T01:09:02Z

@jasperjiaguo Capturing our offline discussion here:

The configuration for enabling this feature works as follows:

PropertyStore will have a tenant-level instance assignment mapping. The name of the ZNode is the tenant name (with instancePartitionType i.e offline,consuming,realtime)
TableConfig.instanceAssignmentConfig will contain "partitionSelector" that denotes whether we have to use MIRROR_SERVER_SET_PARTITION_SELECTOR or not.
We additionally overload instancePartitionsMap in tableConfig with detail about the Znode where the tenant-level instance assignment mapping (1) is stored.

There seems to be a few things we can simplify here:
(1) Remove the additional level of redirection i.e populating instancePartitionsMap can be avoided:

Each table belongs to a specific server tenant
Each tenant has a dedicated Zk node in PropertyStore that stores the instance assignment mapping.
So, given a table, we should automatically be able to figure out the corresponding the tenant-level instance assignment mapping from Zk.

(or)

(2) If we need this extra detail, we should probably avoid overloading the already existing tableConfig.instancePartitionsMap with this mapping detail (even though they are mutually exclusive) and have a new field for it

@jasperjiaguo is going to look at the possibility for this. If the change is non-trivial, it's better left as a TODO.

.../pinot/controller/helix/core/assignment/instance/PreConfiguredInstancePartitionSelector.java

...t-common/src/main/java/org/apache/pinot/common/assignment/InstanceAssignmentConfigUtils.java

...ller/src/main/java/org/apache/pinot/controller/api/resources/PinotTenantRestletResource.java

...inot/controller/helix/core/assignment/instance/MirrorServerSetInstancePartitionSelector.java

...ntroller/src/main/java/org/apache/pinot/controller/helix/core/PinotHelixResourceManager.java

somandal

thanks for the detailed tests!

somandal · 2023-10-28T00:31:01Z

...ava/org/apache/pinot/controller/helix/core/assignment/instance/InstanceAssignmentDriver.java

+        assignmentConfig, instanceConfigs, existingInstancePartitions, null);
+  }
+
+  public InstancePartitions assignInstances(InstancePartitionsType instancePartitionsType,


nit: can the existing assignInstances(InstancePartitionsType instancePartitionsType, List<InstanceConfig> instanceConfigs, @Nullable InstancePartitions existingInstancePartitions) call this function since the remaining setup is the same? just to avoid too much code duplication

...ller/src/main/java/org/apache/pinot/controller/api/resources/PinotTenantRestletResource.java

walterddr · 2023-10-31T19:28:30Z

this PR broken ColocatedJoinQuickStart can we take a look CC @jasperjiaguo @siddharthteotia

walterddr

FYI error:

2023/10/31 12:27:38.097 ERROR [TableConfigsRestletResource] [grizzly-http-server-10] null
java.lang.NullPointerException: null
	at org.apache.pinot.common.assignment.InstanceAssignmentConfigUtils.isMirrorServerSetAssignment(InstanceAssignmentConfigUtils.java:132) ~[classes/:?]
	at org.apache.pinot.controller.helix.core.PinotHelixResourceManager.assignInstances(PinotHelixResourceManager.java:1757) ~[classes/:?]
	at org.apache.pinot.controller.helix.core.PinotHelixResourceManager.addTable(PinotHelixResourceManager.java:1595) ~[classes/:?]
	at org.apache.pinot.controller.api.resources.TableConfigsRestletResource.addConfig(TableConfigsRestletResource.java:227) ~[classes/:?]
	at jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method) ~[?:?]
	at jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) ~[?:?]
	at jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) ~[?:?]
	at java.lang.reflect.Method.invoke(Method.java:566) ~[?:?]
	at org.glassfish.jersey.server.model.internal.ResourceMethodInvocationHandlerFactory.lambda$static$0(ResourceMethodInvocationHandlerFactory.java:52) ~[jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.model.internal.AbstractJavaResourceMethodDispatcher$1.run(AbstractJavaResourceMethodDispatcher.java:134) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.model.internal.AbstractJavaResourceMethodDispatcher.invoke(AbstractJavaResourceMethodDispatcher.java:177) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.model.internal.JavaResourceMethodDispatcherProvider$TypeOutInvoker.doDispatch(JavaResourceMethodDispatcherProvider.java:219) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.model.internal.AbstractJavaResourceMethodDispatcher.dispatch(AbstractJavaResourceMethodDispatcher.java:81) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.model.ResourceMethodInvoker.invoke(ResourceMethodInvoker.java:478) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.model.ResourceMethodInvoker.apply(ResourceMethodInvoker.java:400) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.model.ResourceMethodInvoker.apply(ResourceMethodInvoker.java:81) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.ServerRuntime$1.run(ServerRuntime.java:256) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.internal.Errors$1.call(Errors.java:248) [jersey-common-2.39.jar:?]
	at org.glassfish.jersey.internal.Errors$1.call(Errors.java:244) [jersey-common-2.39.jar:?]
	at org.glassfish.jersey.internal.Errors.process(Errors.java:292) [jersey-common-2.39.jar:?]
	at org.glassfish.jersey.internal.Errors.process(Errors.java:274) [jersey-common-2.39.jar:?]
	at org.glassfish.jersey.internal.Errors.process(Errors.java:244) [jersey-common-2.39.jar:?]
	at org.glassfish.jersey.process.internal.RequestScope.runInScope(RequestScope.java:265) [jersey-common-2.39.jar:?]
	at org.glassfish.jersey.server.ServerRuntime.process(ServerRuntime.java:235) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.server.ApplicationHandler.handle(ApplicationHandler.java:684) [jersey-server-2.39.jar:?]
	at org.glassfish.jersey.grizzly2.httpserver.GrizzlyHttpContainer.service(GrizzlyHttpContainer.java:356) [jersey-container-grizzly2-http-2.39.jar:?]
	at org.glassfish.grizzly.http.server.HttpHandler$1.run(HttpHandler.java:200) [grizzly-http-server-2.4.4.jar:2.4.4]
	at org.glassfish.grizzly.threadpool.AbstractThreadPool$Worker.doWork(AbstractThreadPool.java:569) [grizzly-framework-2.4.4.jar:2.4.4]
	at org.glassfish.grizzly.threadpool.AbstractThreadPool$Worker.run(AbstractThreadPool.java:549) [grizzly-framework-2.4.4.jar:2.4.4]
	at java.lang.Thread.run(Thread.java:829) [?:?]
2023/10/31 12:27:38.105 INFO [AddTableCommand] [main] {"code":500,"error":null}
java.lang.RuntimeException: Unable to create offline table - userGroups from schema file [/var/folders/bq/f8s34x8163v0609ry22yqkf40000gn/T/1698780410918/userGroups/userGroups_schema.json] and table conf file [/var/folders/bq/f8s34x8163v0609ry22yqkf40000gn/T/1698780410918/userGroups/userGroups_offline_table_config.json].
	at org.apache.pinot.tools.BootstrapTableTool.bootstrapOfflineTable(BootstrapTableTool.java:191)
	at org.apache.pinot.tools.BootstrapTableTool.execute(BootstrapTableTool.java:104)
	at org.apache.pinot.tools.admin.command.QuickstartRunner.bootstrapTable(QuickstartRunner.java:232)
	at org.apache.pinot.tools.Quickstart.execute(Quickstart.java:85)
	at org.apache.pinot.tools.admin.command.QuickStartCommand.execute(QuickStartCommand.java:178)

walterddr · 2023-10-31T19:42:53Z

should be fixed by #11915

jasperjiaguo · 2023-11-01T20:28:05Z

should be fixed by #11915

Thanks for the fix!

jasperjiaguo force-pushed the PreConfiguredInstancePartitionSelector branch 3 times, most recently from b52d864 to 4d822e8 Compare September 14, 2023 21:41

jasperjiaguo marked this pull request as ready for review September 14, 2023 21:42

jasperjiaguo force-pushed the PreConfiguredInstancePartitionSelector branch from 4d822e8 to 745e6b0 Compare September 14, 2023 21:55

add tenant partition map rest resource

60877d1

add tenant partition map rest resource fix logic refactor and fix some logic pre-configuration based assignment

jasperjiaguo force-pushed the PreConfiguredInstancePartitionSelector branch from 745e6b0 to 60877d1 Compare October 19, 2023 23:10

jasperjiaguo added 4 commits October 19, 2023 16:13

checkstyle

1c66e5f

checkstyle

19282ce

checkstyle

f77e90b

Trigger Test

0df73b6

Jackie-Jiang added feature Configuration Config changes (addition/deletion/change in behavior) labels Oct 24, 2023

vvivekiyer reviewed Oct 24, 2023

View reviewed changes

Address comments

eb66f9f

vvivekiyer reviewed Oct 25, 2023

View reviewed changes

jasperjiaguo added 7 commits October 24, 2023 18:26

refactored branches

b4cccad

address comments

37d7fe1

address comments

e8587c5

address comments

764c8fc

Change the shuffling logic to be compatible with future changes

9b5c4a9

add test cases

ad941aa

address comments

db49349

vvivekiyer reviewed Oct 26, 2023

View reviewed changes

address comments

dab3299

jasperjiaguo added 2 commits October 25, 2023 20:14

address comments

c646567

change table validation logic

367af71

jasperjiaguo force-pushed the PreConfiguredInstancePartitionSelector branch from 208b079 to 367af71 Compare October 26, 2023 22:41

jasperjiaguo added 2 commits October 26, 2023 16:34

Trigger Test

b044af0

update shuffling logic

545e7d0

somandal reviewed Oct 27, 2023

View reviewed changes

...ntroller/src/main/java/org/apache/pinot/controller/helix/core/PinotHelixResourceManager.java Outdated Show resolved Hide resolved

add more logging

8479b70

somandal approved these changes Oct 28, 2023

View reviewed changes

siddharthteotia approved these changes Oct 30, 2023

View reviewed changes

siddharthteotia merged commit 2e37f2c into apache:master Oct 30, 2023

walterddr reviewed Oct 31, 2023

View reviewed changes

walterddr mentioned this pull request Oct 31, 2023

[hotfix] avoid npe when checking mirror server set assignment #11915

Merged

Conversation

jasperjiaguo commented Sep 12, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov-commenter commented Sep 12, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Jackie-Jiang commented Sep 17, 2023

Uh oh!

Jackie-Jiang commented Oct 24, 2023

Uh oh!

vvivekiyer left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

siddharthteotia commented Oct 24, 2023

Uh oh!

vvivekiyer commented Oct 25, 2023

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

somandal left a comment

Choose a reason for hiding this comment

Uh oh!

somandal Oct 28, 2023

Choose a reason for hiding this comment

Uh oh!

Uh oh!

walterddr commented Oct 31, 2023

Uh oh!

walterddr left a comment

Choose a reason for hiding this comment

Uh oh!

walterddr commented Oct 31, 2023

Uh oh!

jasperjiaguo commented Nov 1, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

jasperjiaguo commented Sep 12, 2023 •

edited

Loading

codecov-commenter commented Sep 12, 2023 •

edited

Loading