You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by "stefankandic (via GitHub)" <gi...@apache.org> on 2024/03/08 13:37:34 UTC
[PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
stefankandic opened a new pull request, #45436:
URL: https://github.com/apache/spark/pull/45436
<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines: https://spark.apache.org/contributing.html
2. Ensure you have added or run the appropriate tests for your PR: https://spark.apache.org/developer-tools.html
3. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP][SPARK-XXXX] Your PR title ...'.
4. Be sure to keep the PR description updated to reflect all changes.
5. Please write your PR title to summarize what this PR proposes.
6. If possible, provide a concise example to reproduce the issue for a faster review.
7. If you want to add a new configuration, please read the guideline first for naming configurations in
'core/src/main/scala/org/apache/spark/internal/config/ConfigEntry.scala'.
8. If you want to add or modify an error type or message, please read the guideline first in
'common/utils/src/main/resources/error/README.md'.
-->
### What changes were proposed in this pull request?
<!--
Please clarify what changes you are proposing. The purpose of this section is to outline the changes and how this PR fixes the issue.
If possible, please consider writing useful notes for better and faster reviews in your PR. See the examples below.
1. If you refactor some codes with changing classes, showing the class hierarchy will help reviewers.
2. If you fix some SQL features, you can provide some references of other DBMSes.
3. If there is design documentation, please add the link.
4. If there is a discussion in the mailing list, please add the link.
-->
Freezing the ICU collator upon creation.
### Why are the changes needed?
<!--
Please clarify why the changes are needed. For instance,
1. If you propose a new API, clarify the use case for a new API.
2. If you fix a bug, you can clarify why it is a bug.
-->
In order to get correct results when getting the collation sort keys using multiple threads. You can read more about collator thread safety [here](https://unicode-org.github.io/icu/userguide/icu/design.html#icu-threading-model-and-open-and-close-model)
### Does this PR introduce _any_ user-facing change?
<!--
Note that it means *any* user-facing change including all aspects such as the documentation fix.
If yes, please clarify the previous behavior and the change this PR proposes - provide the console output, description and/or an example to show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change compared to the released Spark versions or within the unreleased branches such as master.
If no, write 'No'.
-->
No
### How was this patch tested?
<!--
If tests were added, say they were added here. Please make sure to add some test cases that check the changes thoroughly including negative and positive cases if possible.
If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future.
If tests were not added, please describe why they were not added and/or why it was difficult to add.
If benchmark tests were added, please run the benchmarks in GitHub Actions for the consistent environment, and the instructions could accord to: https://spark.apache.org/developer-tools.html#github-workflow-benchmarks.
-->
New unti test
### Was this patch authored or co-authored using generative AI tooling?
<!--
If generative AI tooling has been used in the process of authoring this patch, please include the
phrase: 'Generated-by: ' followed by the name of the tool and its version.
If no, write 'No'.
Please refer to the [ASF Generative Tooling Guidance](https://www.apache.org/legal/generative-tooling.html) for details.
-->
no
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "cloud-fan (via GitHub)" <gi...@apache.org>.
cloud-fan commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1519395144
##########
common/unsafe/src/main/java/org/apache/spark/sql/catalyst/util/CollationFactory.java:
##########
@@ -138,11 +138,13 @@ public Collation(
collationTable[2] = new Collation(
"UNICODE", Collator.getInstance(ULocale.ROOT), "153.120.0.0", true);
collationTable[2].collator.setStrength(Collator.TERTIARY);
+ collationTable[2].collator.freeze();
Review Comment:
they should use one buffer per thread... Anyway this is out of our control and calling `freeze` LGTM
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "stefankandic (via GitHub)" <gi...@apache.org>.
stefankandic commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1519361380
##########
common/unsafe/src/main/java/org/apache/spark/sql/catalyst/util/CollationFactory.java:
##########
@@ -138,11 +138,13 @@ public Collation(
collationTable[2] = new Collation(
"UNICODE", Collator.getInstance(ULocale.ROOT), "153.120.0.0", true);
collationTable[2].collator.setStrength(Collator.TERTIARY);
+ collationTable[2].collator.freeze();
Review Comment:
@cloud-fan collator uses a buffer while writing collation keys, freezing it makes this operation safe by using a reentrant lock around it ([source](https://github.com/unicode-org/icu/blob/main/icu4j/main/collate/src/main/java/com/ibm/icu/text/RuleBasedCollator.java#L1831))
this of course raises performance issues which we should probably discuss, because now only one thread can generate sort keys
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "HyukjinKwon (via GitHub)" <gi...@apache.org>.
HyukjinKwon closed pull request #45436: [SPARK-47327][SQL] Fix thread safety issue in ICU Collator
URL: https://github.com/apache/spark/pull/45436
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "HyukjinKwon (via GitHub)" <gi...@apache.org>.
HyukjinKwon commented on PR #45436:
URL: https://github.com/apache/spark/pull/45436#issuecomment-1989650161
Merged to master.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "dbatomic (via GitHub)" <gi...@apache.org>.
dbatomic commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1519347155
##########
sql/core/src/test/scala/org/apache/spark/sql/CollationSuite.scala:
##########
@@ -438,6 +439,19 @@ class CollationSuite extends DatasourceV2SQLBase with AdaptiveSparkPlanHelper {
}
}
+ test("test concurrently generating collation keys") {
Review Comment:
Can this test go to `CollationFactorySuite`?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "uros-db (via GitHub)" <gi...@apache.org>.
uros-db commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1518058827
##########
sql/core/src/test/scala/org/apache/spark/sql/CollationSuite.scala:
##########
@@ -438,6 +438,39 @@ class CollationSuite extends DatasourceV2SQLBase with AdaptiveSparkPlanHelper {
}
}
+ test("test concurrently running aggregates") {
+ // generating ICU sort keys is not thread-safe by default so this should fail
+ // if we don't handle the concurrency properly on Collator level
+
+ for (_ <- 1 to 100) {
Review Comment:
just wondering: have you tried 1000, 10000, etc. or is 100 proven to be sufficient?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "MaxGekk (via GitHub)" <gi...@apache.org>.
MaxGekk commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1518208382
##########
sql/core/src/test/scala/org/apache/spark/sql/CollationSuite.scala:
##########
@@ -438,6 +438,39 @@ class CollationSuite extends DatasourceV2SQLBase with AdaptiveSparkPlanHelper {
}
}
+ test("test concurrently running aggregates") {
+ // generating ICU sort keys is not thread-safe by default so this should fail
+ // if we don't handle the concurrency properly on Collator level
+
+ for (_ <- 1 to 100) {
+ Seq(
+ ("ucs_basic", Seq("AAA", "aaa"), Seq(Row(1, "AAA"), Row(1, "aaa"))),
+ ("ucs_basic", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("ucs_basic", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb"))),
+ ("ucs_basic_lcase", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("ucs_basic_lcase", Seq("AAA", "aaa"), Seq(Row(2, "AAA"))),
+ ("ucs_basic_lcase", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb"))),
+ ("unicode", Seq("AAA", "aaa"), Seq(Row(1, "AAA"), Row(1, "aaa"))),
+ ("unicode", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("unicode", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb"))),
+ ("unicode_CI", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("unicode_CI", Seq("AAA", "aaa"), Seq(Row(2, "AAA"))),
+ ("unicode_CI", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb")))
+ ).foreach {
Review Comment:
How this sequential execution proofs the fix? Maybe use a parallel collection?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "cloud-fan (via GitHub)" <gi...@apache.org>.
cloud-fan commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1519141471
##########
common/unsafe/src/main/java/org/apache/spark/sql/catalyst/util/CollationFactory.java:
##########
@@ -138,11 +138,13 @@ public Collation(
collationTable[2] = new Collation(
"UNICODE", Collator.getInstance(ULocale.ROOT), "153.120.0.0", true);
collationTable[2].collator.setStrength(Collator.TERTIARY);
+ collationTable[2].collator.freeze();
Review Comment:
This is interesting, I thought `Collator` is just holding a bunch of functions, but it has mutable states?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "stefankandic (via GitHub)" <gi...@apache.org>.
stefankandic commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1518248868
##########
sql/core/src/test/scala/org/apache/spark/sql/CollationSuite.scala:
##########
@@ -438,6 +438,39 @@ class CollationSuite extends DatasourceV2SQLBase with AdaptiveSparkPlanHelper {
}
}
+ test("test concurrently running aggregates") {
+ // generating ICU sort keys is not thread-safe by default so this should fail
+ // if we don't handle the concurrency properly on Collator level
+
+ for (_ <- 1 to 100) {
+ Seq(
+ ("ucs_basic", Seq("AAA", "aaa"), Seq(Row(1, "AAA"), Row(1, "aaa"))),
+ ("ucs_basic", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("ucs_basic", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb"))),
+ ("ucs_basic_lcase", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("ucs_basic_lcase", Seq("AAA", "aaa"), Seq(Row(2, "AAA"))),
+ ("ucs_basic_lcase", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb"))),
+ ("unicode", Seq("AAA", "aaa"), Seq(Row(1, "AAA"), Row(1, "aaa"))),
+ ("unicode", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("unicode", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb"))),
+ ("unicode_CI", Seq("aaa", "aaa"), Seq(Row(2, "aaa"))),
+ ("unicode_CI", Seq("AAA", "aaa"), Seq(Row(2, "AAA"))),
+ ("unicode_CI", Seq("aaa", "bbb"), Seq(Row(1, "aaa"), Row(1, "bbb")))
+ ).foreach {
Review Comment:
that's a very good point, the test would fail simply because it would be ran 100 times and at least one of those execution would have a data race - i improved it now to just call `getCollationKey` in a parallel for each and not really on spark's execution of the aggregate query
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "uros-db (via GitHub)" <gi...@apache.org>.
uros-db commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1518055476
##########
common/unsafe/src/main/java/org/apache/spark/sql/catalyst/util/CollationFactory.java:
##########
@@ -138,11 +138,13 @@ public Collation(
collationTable[2] = new Collation(
"UNICODE", Collator.getInstance(ULocale.ROOT), "153.120.0.0", true);
collationTable[2].collator.setStrength(Collator.TERTIARY);
+ collationTable[2].collator.freeze();
Review Comment:
I've never seen "freeze" before, so I'm wondering how this affects us.
As I understood from the ICU docs: Once frozen, an object can never be unfrozen, so it is thread-safe from that point onward.
So what are the drawbacks of this apprach?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "dbatomic (via GitHub)" <gi...@apache.org>.
dbatomic commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1519437339
##########
common/unsafe/src/main/java/org/apache/spark/sql/catalyst/util/CollationFactory.java:
##########
@@ -138,11 +138,13 @@ public Collation(
collationTable[2] = new Collation(
"UNICODE", Collator.getInstance(ULocale.ROOT), "153.120.0.0", true);
collationTable[2].collator.setStrength(Collator.TERTIARY);
+ collationTable[2].collator.freeze();
Review Comment:
Yeah, as soon as we get benchmarks working we should revisit this decision.
One option that we also prototyped is to keep `Collator` in `ThreadLocal` fields, which also solved the problem. But `freeze` is a bit cleaner and we don't have microbenchmarks yet so we can't make data driven decision at this point.
LGTM.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "stefankandic (via GitHub)" <gi...@apache.org>.
stefankandic commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1519361380
##########
common/unsafe/src/main/java/org/apache/spark/sql/catalyst/util/CollationFactory.java:
##########
@@ -138,11 +138,13 @@ public Collation(
collationTable[2] = new Collation(
"UNICODE", Collator.getInstance(ULocale.ROOT), "153.120.0.0", true);
collationTable[2].collator.setStrength(Collator.TERTIARY);
+ collationTable[2].collator.freeze();
Review Comment:
@cloud-fan collator uses a buffer while writing collation keys, freezing it makes this operation safe by using a reentrant lock around it ([source](https://github.com/unicode-org/icu/blob/main/icu4j/main/collate/src/main/java/com/ibm/icu/text/RuleBasedCollator.java#L1831))
this of course raises performance issues which we should probably discuss, because now we can't generate sort keys in parallel on a single collator
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "stefankandic (via GitHub)" <gi...@apache.org>.
stefankandic commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1519362819
##########
sql/core/src/test/scala/org/apache/spark/sql/CollationSuite.scala:
##########
@@ -438,6 +439,19 @@ class CollationSuite extends DatasourceV2SQLBase with AdaptiveSparkPlanHelper {
}
}
+ test("test concurrently generating collation keys") {
Review Comment:
it could but I decided to put it here because it would require adding a new dependency for parallel collections which I'd like to avoid
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org
Re: [PR] [SPARK-47327][SQL] Fix thread safety issue in ICU Collator [spark]
Posted by "uros-db (via GitHub)" <gi...@apache.org>.
uros-db commented on code in PR #45436:
URL: https://github.com/apache/spark/pull/45436#discussion_r1518059723
##########
sql/core/src/test/scala/org/apache/spark/sql/CollationSuite.scala:
##########
@@ -438,6 +438,39 @@ class CollationSuite extends DatasourceV2SQLBase with AdaptiveSparkPlanHelper {
}
}
+ test("test concurrently running aggregates") {
Review Comment:
on another note, do we have some kind of _truly_ multithread tests?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org