This change introduces two optimizations to help speed up generation
of listing data when parsing events logs.
The first one allows the parser to be stopped when enough data to
create the listing entry has been read. This is currently the start
event plus environment info, to capture UI ACLs. If the end event is
needed, the code will skip to the end of the log to try to find that
information, instead of parsing the whole log file.
Unfortunately this works better with uncompressed logs. Skipping bytes
on compressed logs only saves the work of parsing lines and some events,
so not a lot of gains are observed.
The second optimization deals with in-progress logs. It works in two
ways: first, it completely avoids parsing the rest of the log for
these apps when enough listing data is read. This, unlike the above,
also speeds things up for compressed logs, since only the very beginning
of the log has to be read.
On top of that, the code that decides whether to re-parse logs to get
updated listing data will ignore in-progress applications until they've
completed.
Both optimizations can be disabled but are enabled by default.
I tested this on some fake event logs to see the effect. I created
500 logs of about 60M each (so ~30G uncompressed; each log was 1.7M
when compressed with zstd). Below, C = completed, IP = in-progress,
the size means the amount of data re-parsed at the end of logs
when necessary.
```
none/C none/IP zstd/C zstd/IP
On / 16k 2s 2s 22s 2s
On / 1m 3s 2s 24s 2s
Off 1.1m 1.1m 26s 24s
```
This was with 4 threads on a single local SSD. As expected from the
previous explanations, there are considerable gains for in-progress
logs, and for uncompressed logs, but not so much when looking at the
full compressed log.
As a side note, I removed the custom code to get the scan time by
creating a file on HDFS; since file mod times are not used to detect
changed logs anymore, local time is enough for the current use of
the SHS.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20952 from vanzin/SPARK-6951.
SPARK-19276 ensured that FetchFailures do not get swallowed by other
layers of exception handling, but it also meant that a killed task could
look like a fetch failure. This is particularly a problem with
speculative execution, where we expect to kill tasks as they are reading
shuffle data. The fix is to ensure that we always check for killed
tasks first.
Added a new unit test which fails before the fix, ran it 1k times to
check for flakiness. Full suite of tests on jenkins.
Author: Imran Rashid <irashid@cloudera.com>
Closes#20987 from squito/SPARK-23816.
## What changes were proposed in this pull request?
The test case JobCancellationSuite."interruptible iterator of shuffle reader" has been flaky because `KillTask` event is handled asynchronously, so it can happen that the semaphore is released but the task is still running.
Actually we only have to check if the total number of processed elements is less than the input elements number, so we know the task get cancelled.
## How was this patch tested?
The new test case still fails without the purposed patch, and succeeded in current master.
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20993 from jiangxb1987/JobCancellationSuite.
## What changes were proposed in this pull request?
This PR avoids possible overflow at an operation `long = (long)(int * int)`. The multiplication of large positive integer values may set one to MSB. This leads to a negative value in long while we expected a positive value (e.g. `0111_0000_0000_0000 * 0000_0000_0000_0010`).
This PR performs long cast before the multiplication to avoid this situation.
## How was this patch tested?
Existing UTs
Author: Kazuaki Ishizaki <ishizaki@jp.ibm.com>
Closes#21002 from kiszk/SPARK-23893.
## What changes were proposed in this pull request?
This PR allows us to use one of several types of `MemoryBlock`, such as byte array, int array, long array, or `java.nio.DirectByteBuffer`. To use `java.nio.DirectByteBuffer` allows to have off heap memory which is automatically deallocated by JVM. `MemoryBlock` class has primitive accessors like `Platform.getInt()`, `Platform.putint()`, or `Platform.copyMemory()`.
This PR uses `MemoryBlock` for `OffHeapColumnVector`, `UTF8String`, and other places. This PR can improve performance of operations involving memory accesses (e.g. `UTF8String.trim`) by 1.8x.
For now, this PR does not use `MemoryBlock` for `BufferHolder` based on cloud-fan's [suggestion](https://github.com/apache/spark/pull/11494#issuecomment-309694290).
Since this PR is a successor of #11494, close#11494. Many codes were ported from #11494. Many efforts were put here. **I think this PR should credit to yzotov.**
This PR can achieve **1.1-1.4x performance improvements** for operations in `UTF8String` or `Murmur3_x86_32`. Other operations are almost comparable performances.
Without this PR
```
OpenJDK 64-Bit Server VM 1.8.0_121-8u121-b13-0ubuntu1.16.04.2-b13 on Linux 4.4.0-22-generic
Intel(R) Xeon(R) CPU E5-2667 v3 3.20GHz
OpenJDK 64-Bit Server VM 1.8.0_121-8u121-b13-0ubuntu1.16.04.2-b13 on Linux 4.4.0-22-generic
Intel(R) Xeon(R) CPU E5-2667 v3 3.20GHz
Hash byte arrays with length 268435487: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Murmur3_x86_32 526 / 536 0.0 131399881.5 1.0X
UTF8String benchmark: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
hashCode 525 / 552 1022.6 1.0 1.0X
substring 414 / 423 1298.0 0.8 1.3X
```
With this PR
```
OpenJDK 64-Bit Server VM 1.8.0_121-8u121-b13-0ubuntu1.16.04.2-b13 on Linux 4.4.0-22-generic
Intel(R) Xeon(R) CPU E5-2667 v3 3.20GHz
Hash byte arrays with length 268435487: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Murmur3_x86_32 474 / 488 0.0 118552232.0 1.0X
UTF8String benchmark: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
hashCode 476 / 480 1127.3 0.9 1.0X
substring 287 / 291 1869.9 0.5 1.7X
```
Benchmark program
```
test("benchmark Murmur3_x86_32") {
val length = 8192 * 32768 + 31
val seed = 42L
val iters = 1 << 2
val random = new Random(seed)
val arrays = Array.fill[MemoryBlock](numArrays) {
val bytes = new Array[Byte](length)
random.nextBytes(bytes)
new ByteArrayMemoryBlock(bytes, Platform.BYTE_ARRAY_OFFSET, length)
}
val benchmark = new Benchmark("Hash byte arrays with length " + length,
iters * numArrays, minNumIters = 20)
benchmark.addCase("HiveHasher") { _: Int =>
var sum = 0L
for (_ <- 0L until iters) {
sum += HiveHasher.hashUnsafeBytesBlock(
arrays(i), Platform.BYTE_ARRAY_OFFSET, length)
}
}
benchmark.run()
}
test("benchmark UTF8String") {
val N = 512 * 1024 * 1024
val iters = 2
val benchmark = new Benchmark("UTF8String benchmark", N, minNumIters = 20)
val str0 = new java.io.StringWriter() { { for (i <- 0 until N) { write(" ") } } }.toString
val s0 = UTF8String.fromString(str0)
benchmark.addCase("hashCode") { _: Int =>
var h: Int = 0
for (_ <- 0L until iters) { h += s0.hashCode }
}
benchmark.addCase("substring") { _: Int =>
var s: UTF8String = null
for (_ <- 0L until iters) { s = s0.substring(N / 2 - 5, N / 2 + 5) }
}
benchmark.run()
}
```
I run [this benchmark program](https://gist.github.com/kiszk/94f75b506c93a663bbbc372ffe8f05de) using [the commit](ee5a79861c). I got the following results:
```
OpenJDK 64-Bit Server VM 1.8.0_151-8u151-b12-0ubuntu0.16.04.2-b12 on Linux 4.4.0-66-generic
Intel(R) Xeon(R) CPU E5-2667 v3 3.20GHz
Memory access benchmarks: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
ByteArrayMemoryBlock get/putInt() 220 / 221 609.3 1.6 1.0X
Platform get/putInt(byte[]) 220 / 236 610.9 1.6 1.0X
Platform get/putInt(Object) 492 / 494 272.8 3.7 0.4X
OnHeapMemoryBlock get/putLong() 322 / 323 416.5 2.4 0.7X
long[] 221 / 221 608.0 1.6 1.0X
Platform get/putLong(long[]) 321 / 321 418.7 2.4 0.7X
Platform get/putLong(Object) 561 / 563 239.2 4.2 0.4X
```
I also run [this benchmark program](https://gist.github.com/kiszk/5fdb4e03733a5d110421177e289d1fb5) for comparing performance of `Platform.copyMemory()`.
```
OpenJDK 64-Bit Server VM 1.8.0_151-8u151-b12-0ubuntu0.16.04.2-b12 on Linux 4.4.0-66-generic
Intel(R) Xeon(R) CPU E5-2667 v3 3.20GHz
Platform copyMemory: Best/Avg Time(ms) Rate(M/s) Per Row(ns) Relative
------------------------------------------------------------------------------------------------
Object to Object 1961 / 1967 8.6 116.9 1.0X
System.arraycopy Object to Object 1917 / 1921 8.8 114.3 1.0X
byte array to byte array 1961 / 1968 8.6 116.9 1.0X
System.arraycopy byte array to byte array 1909 / 1937 8.8 113.8 1.0X
int array to int array 1921 / 1990 8.7 114.5 1.0X
double array to double array 1918 / 1923 8.7 114.3 1.0X
Object to byte array 1961 / 1967 8.6 116.9 1.0X
Object to short array 1965 / 1972 8.5 117.1 1.0X
Object to int array 1910 / 1915 8.8 113.9 1.0X
Object to float array 1971 / 1978 8.5 117.5 1.0X
Object to double array 1919 / 1944 8.7 114.4 1.0X
byte array to Object 1959 / 1967 8.6 116.8 1.0X
int array to Object 1961 / 1970 8.6 116.9 1.0X
double array to Object 1917 / 1924 8.8 114.3 1.0X
```
These results show three facts:
1. According to the second/third or sixth/seventh results in the first experiment, if we use `Platform.get/putInt(Object)`, we achieve more than 2x worse performance than `Platform.get/putInt(byte[])` with concrete type (i.e. `byte[]`).
2. According to the second/third or fourth/fifth/sixth results in the first experiment, the fastest way to access an array element on Java heap is `array[]`. **Cons of `array[]` is that it is not possible to support unaligned-8byte access.**
3. According to the first/second/third or fourth/sixth/seventh results in the first experiment, `getInt()/putInt() or getLong()/putLong()` in subclasses of `MemoryBlock` can achieve comparable performance to `Platform.get/putInt()` or `Platform.get/putLong()` with concrete type (second or sixth result). There is no overhead regarding virtual call.
4. According to results in the second experiment, for `Platform.copy()`, to pass `Object` can achieve the same performance as to pass any type of primitive array as source or destination.
5. According to second/fourth results in the second experiment, `Platform.copy()` can achieve the same performance as `System.arrayCopy`. **It would be good to use `Platform.copy()` since `Platform.copy()` can take any types for src and dst.**
We are incrementally replace `Platform.get/putXXX` with `MemoryBlock.get/putXXX`. This is because we have two advantages.
1) Achieve better performance due to having a concrete type for an array.
2) Use simple OO design instead of passing `Object`
It is easy to use `MemoryBlock` in `InternalRow`, `BufferHolder`, `TaskMemoryManager`, and others that are already abstracted. It is not easy to use `MemoryBlock` in utility classes related to hashing or others.
Other candidates are
- UnsafeRow, UnsafeArrayData, UnsafeMapData, SpecificUnsafeRowJoiner
- UTF8StringBuffer
- BufferHolder
- TaskMemoryManager
- OnHeapColumnVector
- BytesToBytesMap
- CachedBatch
- classes for hash
- others.
## How was this patch tested?
Added `UnsafeMemoryAllocator`
Author: Kazuaki Ishizaki <ishizaki@jp.ibm.com>
Closes#19222 from kiszk/SPARK-10399.
## What changes were proposed in this pull request?
Address https://github.com/apache/spark/pull/20924#discussion_r177987175, show block manager id when remove RDD/Broadcast fails.
## How was this patch tested?
N/A
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20960 from jiangxb1987/bmid.
These tests can fail with a timeout if the remote repos are not responding,
or slow. The tests don't need anything from those repos, so use an empty
ivy config file to avoid setting up the defaults.
The tests are passing reliably for me locally now, and failing more often
than not today without this change since http://dl.bintray.com/spark-packages/maven
doesn't seem to be loading from my machine.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20916 from vanzin/SPARK-19964.
## What changes were proposed in this pull request?
Address https://github.com/apache/spark/pull/20449#discussion_r172414393, If `resultIter` is already a `InterruptibleIterator`, don't double wrap it.
## How was this patch tested?
Existing tests.
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20920 from jiangxb1987/SPARK-23040.
## What changes were proposed in this pull request?
It may be get `spark.shuffle.service.port` from 9745ec3a61/core/src/main/scala/org/apache/spark/deploy/SparkHadoopUtil.scala (L459)
Therefore, the client configuration `spark.shuffle.service.port` does not working unless the configuration is `spark.hadoop.spark.shuffle.service.port`.
- This configuration is not working:
```
bin/spark-sql --master yarn --conf spark.shuffle.service.port=7338
```
- This configuration works:
```
bin/spark-sql --master yarn --conf spark.hadoop.spark.shuffle.service.port=7338
```
This PR fix this issue.
## How was this patch tested?
It's difficult to carry out unit testing. But I've tested it manually.
Author: Yuming Wang <yumwang@ebay.com>
Closes#20785 from wangyum/SPARK-23640.
## What changes were proposed in this pull request?
In SparkSQLCLI, SessionState generates before SparkContext instantiating. When we use --proxy-user to impersonate, it's unable to initializing a metastore client to talk to the secured metastore for no kerberos ticket.
This PR use real user ugi to obtain token for owner before talking to kerberized metastore.
## How was this patch tested?
Manually verified with kerberized hive metasotre / hdfs.
Author: Kent Yao <yaooqinn@hotmail.com>
Closes#20784 from yaooqinn/SPARK-23639.
## What changes were proposed in this pull request?
Changed `LauncherBackend` `set` method so that it checks if the connection is open or not before writing to it (uses `isConnected`).
## How was this patch tested?
None
Author: Sahil Takiar <stakiar@cloudera.com>
Closes#20893 from sahilTakiar/master.
… with dynamic allocation
## What changes were proposed in this pull request?
ignore errors when you are waiting for a broadcast.unpersist. This is handling it the same way as doing rdd.unpersist in https://issues.apache.org/jira/browse/SPARK-22618
## How was this patch tested?
Patch was tested manually against a couple jobs that exhibit this behavior, with the change the application no longer dies due to this and just prints the warning.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Thomas Graves <tgraves@unharmedunarmed.corp.ne1.yahoo.com>
Closes#20924 from tgravescs/SPARK-23806.
This change basically rewrites the security documentation so that it's
up to date with new features, more correct, and more complete.
Because security is such an important feature, I chose to move all the
relevant configuration documentation to the security page, instead of
having them peppered all over the place in the configuration page. This
allows an almost one-stop shop for security configuration in Spark. The
only exceptions are some YARN-specific minor features which I left in
the YARN page.
I also re-organized the page's topics, since they didn't make a lot of
sense. You had kerberos features described inside paragraphs talking
about UI access control, and other oddities. It should be easier now
to find information about specific Spark security features. I also
enabled TOCs for both the Security and YARN pages, since that makes it
easier to see what is covered.
I removed most of the comments from the SecurityManager javadoc since
they just replicated information in the security doc, with different
levels of out-of-dateness.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20742 from vanzin/SPARK-23572.
This particular test assumed that Hadoop libraries did not support
http as a file system. Hadoop 2.9 does, so the test failed. The test
now forces a non-existent implementation for the http fs, which
forces the expected error.
There were also a couple of other issues in the same test: SparkSubmit
arguments in the wrong order, and the wrong check later when asserting,
which was being masked by the previous issues.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20895 from vanzin/SPARK-23787.
## What changes were proposed in this pull request?
Fixes SPARK-23759 by moving connector.start() after connector.setHost()
Problem was created due connector.setHost(hostName) call was after connector.start()
## How was this patch tested?
Patch was tested after build and deployment. This patch requires SPARK_LOCAL_IP environment variable to be set on spark-env.sh
Author: bag_of_tricks <falbani@hortonworks.com>
Closes#20883 from felixalbani/SPARK-23759.
## What changes were proposed in this pull request?
We re-enabled the Scalastyle checker on a line of code. It was previously disabled, but it does not violate any of the rules. So there's no reason to disable the Scalastyle checker here.
## How was this patch tested?
We tested this by running `build/mvn scalastyle:check` after removing the comments that disable the checker. This check passed with no errors or warnings for Spark Core
```
[INFO]
[INFO] ------------------------------------------------------------------------
[INFO] Building Spark Project Core 2.4.0-SNAPSHOT
[INFO] ------------------------------------------------------------------------
[INFO]
[INFO] --- scalastyle-maven-plugin:1.0.0:check (default-cli) spark-core_2.11 ---
Saving to outputFile=<path to local dir>/spark/core/target/scalastyle-output.xml
Processed 485 file(s)
Found 0 errors
Found 0 warnings
Found 0 infos
```
We did not run all tests (with `dev/run-tests`) since this Scalastyle check seemed sufficient.
## Co-contributors:
chialun-yeh
Hrayo712
vpourquie
Author: arucard21 <arucard21@gmail.com>
Closes#20880 from arucard21/scalastyle_util.
Currently, the Spark AM relies on the initial set of tokens created by
the submission client to be able to talk to HDFS and other services that
require delegation tokens. This means that after those tokens expire, a
new AM will fail to start (e.g. when there is an application failure and
re-attempts are enabled).
This PR makes it so that the first thing the AM does when the user provides
a principal and keytab is to create new delegation tokens for use. This
makes sure that the AM can be started irrespective of how old the original
token set is. It also allows all of the token management to be done by the
AM - there is no need for the submission client to set configuration values
to tell the AM when to renew tokens.
Note that even though in this case the AM will not be using the delegation
tokens created by the submission client, those tokens still need to be provided
to YARN, since they are used to do log aggregation.
To be able to re-use the code in the AMCredentialRenewal for the above
purposes, I refactored that class a bit so that it can fetch tokens into
a pre-defined UGI, insted of always logging in.
Another issue with re-attempts is that, after the fix that allows the AM
to restart correctly, new executors would get confused about when to
update credentials, because the credential updater used the update time
initially set up by the submission code. This could make the executor
fail to update credentials in time, since that value would be very out
of date in the situation described in the bug.
To fix that, I changed the YARN code to use the new RPC-based mechanism
for distributing tokens to executors. This allowed the old credential
updater code to be removed, and a lot of code in the renewer to be
simplified.
I also made two currently hardcoded values (the renewal time ratio, and
the retry wait) configurable; while this probably never needs to be set
by anyone in a production environment, it helps with testing; that's also
why they're not documented.
Tested on real cluster with a specially crafted application to test this
functionality: checked proper access to HDFS, Hive and HBase in cluster
mode with token renewal on and AM restarts. Tested things still work in
client mode too.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20657 from vanzin/SPARK-23361.
Firstly, glob resolution will not result in swallowing the remote name part (that is preceded by the `#` sign) in case of `--files` or `--archives` options
Moreover in the special case of multiple resolutions when the remote naming does not make sense and error is returned.
Enhanced current test and wrote additional test for the error case
Author: Mihaly Toth <misutoth@gmail.com>
Closes#20853 from misutoth/glob-with-remote-name.
## What changes were proposed in this pull request?
Removal of the init-container for downloading remote dependencies. Built off of the work done by vanzin in an attempt to refactor driver/executor configuration elaborated in [this](https://issues.apache.org/jira/browse/SPARK-22839) ticket.
## How was this patch tested?
This patch was tested with unit and integration tests.
Author: Ilan Filonenko <if56@cornell.edu>
Closes#20669 from ifilonenko/remove-init-container.
## What changes were proposed in this pull request?
Minor modification.Comment below is not right.
```
/**
* Adds a shutdown hook with the given priority. Hooks with lower priority values run
* first.
*
* param hook The code to run during shutdown.
* return A handle that can be used to unregister the shutdown hook.
*/
def addShutdownHook(priority: Int)(hook: () => Unit): AnyRef = {
shutdownHooks.add(priority, hook)
}
```
## How was this patch tested?
UT
Author: zhoukang <zhoukang199191@gmail.com>
Closes#20845 from caneGuy/zhoukang/fix-shutdowncomment.
## What changes were proposed in this pull request?
With SPARK-20236, `FileCommitProtocol.instantiate()` looks for a three argument constructor, passing in the `dynamicPartitionOverwrite` parameter. If there is no such constructor, it falls back to the classic two-arg one.
When `InsertIntoHadoopFsRelationCommand` passes down that `dynamicPartitionOverwrite` flag `to FileCommitProtocol.instantiate(`), it assumes that the instantiated protocol supports the specific requirements of dynamic partition overwrite. It does not notice when this does not hold, and so the output generated may be incorrect.
This patch changes `FileCommitProtocol.instantiate()` so when `dynamicPartitionOverwrite == true`, it requires the protocol implementation to have a 3-arg constructor. Classic two arg constructors are supported when it is false.
Also it adds some debug level logging for anyone trying to understand what's going on.
## How was this patch tested?
Unit tests verify that
* classes with only 2-arg constructor cannot be used with dynamic overwrite
* classes with only 2-arg constructor can be used without dynamic overwrite
* classes with 3 arg constructors can be used with both.
* the fallback to any two arg ctor takes place after the attempt to load the 3-arg ctor,
* passing in invalid class types fail as expected (regression tests on expected behavior)
Author: Steve Loughran <stevel@hortonworks.com>
Closes#20824 from steveloughran/stevel/SPARK-23683-protocol-instantiate.
## What changes were proposed in this pull request?
SHS is using a relative path for the REST API call to get the list of the application is a relative path call. In case of the SHS being consumed through a proxy, it can be an issue if the path doesn't end with a "/".
Therefore, we should use an absolute path for the REST call as it is done for all the other resources.
## How was this patch tested?
manual tests
Before the change:
After the change:
Author: Marco Gaido <marcogaido91@gmail.com>
Closes#20794 from mgaido91/SPARK-23644.
Jetty handlers are dynamically attached/detached while SHS is running. But the attach and detach operations might be taking place at the same time due to the async in load/clear in Guava Cache.
## What changes were proposed in this pull request?
Add synchronization between attachSparkUI and detachSparkUI in SHS.
## How was this patch tested?
With this patch, the jetty handlers missing issue never happens again in our production cluster SHS.
Author: Ye Zhou <yezhou@linkedin.com>
Closes#20744 from zhouyejoe/SPARK-23608.
Added/corrected scaladoc for isZero on the DoubleAccumulator, CollectionAccumulator, and LongAccumulator subclasses of AccumulatorV2, particularly noting where there are requirements in addition to having a value of zero in order to return true.
## What changes were proposed in this pull request?
Three scaladoc comments are updated in AccumulatorV2.scala
No changes outside of comment blocks were made.
## How was this patch tested?
Running "sbt unidoc", fixing style errors found, and reviewing the resulting local scaladoc in firefox.
Author: smallory <s.mallory@gmail.com>
Closes#20790 from smallory/patch-1.
This change restores functionality that was inadvertently removed as part
of the fix for SPARK-22372.
Also modified an existing unit test to make sure the feature works as intended.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20776 from vanzin/SPARK-23630.
## What changes were proposed in this pull request?
I propose to replace `'\n'` by the `<br>` tag in generated html of thread dump page. The `<br>` tag will split thread lines in more reliable way. For now it could look like on
<img width="1265" alt="the screen shot" src="https://user-images.githubusercontent.com/1580697/37118202-bcd98fc0-2253-11e8-9e61-c2f946869ee0.png">
if the html is proxied and `'\n'` is replaced by another whitespace. The changes allow to more easily read and copy stack traces.
## How was this patch tested?
I tested it manually by checking the thread dump page and its source.
Author: Maxim Gekk <maxim.gekk@databricks.com>
Closes#20762 from MaxGekk/br-thread-dump.
A few different things going on:
- Remove unused methods.
- Move JSON methods to the only class that uses them.
- Move test-only methods to TestUtils.
- Make getMaxResultSize() a config constant.
- Reuse functionality from existing libraries (JRE or JavaUtils) where possible.
The change also includes changes to a few tests to call `Utils.createTempFile` correctly,
so that temp dirs are created under the designated top-level temp dir instead of
potentially polluting git index.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20706 from vanzin/SPARK-23550.
These options were used to configure the built-in JRE SSL libraries
when downloading files from HTTPS servers. But because they were also
used to set up the now (long) removed internal HTTPS file server,
their default configuration chose convenience over security by having
overly lenient settings.
This change removes the configuration options that affect the JRE SSL
libraries. The JRE trust store can still be configured via system
properties (or globally in the JRE security config). The only lost
functionality is not being able to disable the default hostname
verifier when using spark-submit, which should be fine since Spark
itself is not using https for any internal functionality anymore.
I also removed the HTTP-related code from the REPL class loader, since
we haven't had a HTTP server for REPL-generated classes for a while.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20723 from vanzin/SPARK-23538.
## What changes were proposed in this pull request?
Before this commit, a non-interruptible iterator is returned if aggregator or ordering is specified.
This commit also ensures that sorter is closed even when task is cancelled(killed) in the middle of sorting.
## How was this patch tested?
Add a unit test in JobCancellationSuite
Author: Xianjin YE <advancedxy@gmail.com>
Closes#20449 from advancedxy/SPARK-23040.
## What changes were proposed in this pull request?
The algorithm in `DefaultPartitionCoalescer.setupGroups` is responsible for picking preferred locations for coalesced partitions. It analyzes the preferred locations of input partitions. It starts by trying to create one partition for each unique location in the input. However, if the the requested number of coalesced partitions is higher that the number of unique locations, it has to pick duplicate locations.
Previously, the duplicate locations would be picked by iterating over the input partitions in order, and copying their preferred locations to coalesced partitions. If the input partitions were clustered by location, this could result in severe skew.
With the fix, instead of iterating over the list of input partitions in order, we pick them at random. It's not perfectly balanced, but it's much better.
## How was this patch tested?
Unit test reproducing the behavior was added.
Author: Ala Luszczak <ala@databricks.com>
Closes#20664 from ala/SPARK-23496.
## What changes were proposed in this pull request?
When the shuffle dependency specifies aggregation ,and `dependency.mapSideCombine=false`, in the map side,there is no need for aggregation and sorting, so we should be able to use serialized sorting.
## How was this patch tested?
Existing unit test
Author: liuxian <liu.xian3@zte.com.cn>
Closes#20576 from 10110346/mapsidecombine.
… cause oom
## What changes were proposed in this pull request?
blockManagerIdCache in BlockManagerId will not remove old values which may cause oom
`val blockManagerIdCache = new ConcurrentHashMap[BlockManagerId, BlockManagerId]()`
Since whenever we apply a new BlockManagerId, it will put into this map.
This patch will use guava cahce for blockManagerIdCache instead.
A heap dump show in [SPARK-23508](https://issues.apache.org/jira/browse/SPARK-23508)
## How was this patch tested?
Exist tests.
Author: zhoukang <zhoukang199191@gmail.com>
Closes#20667 from caneGuy/zhoukang/fix-history.
As suggested in #20651, the code is very redundant in `AllStagesPage` and modifying it is a copy-and-paste work. We should avoid such a pattern, which is error prone, and have a cleaner solution which avoids code redundancy.
existing UTs
Author: Marco Gaido <marcogaido91@gmail.com>
Closes#20663 from mgaido91/SPARK-23475_followup.
The ExecutorAllocationManager should not adjust the target number of
executors when killing idle executors, as it has already adjusted the
target number down based on the task backlog.
The name `replace` was misleading with DynamicAllocation on, as the target number
of executors is changed outside of the call to `killExecutors`, so I adjusted that name. Also separated out the logic of `countFailures` as you don't always want that tied to `replace`.
While I was there I made two changes that weren't directly related to this:
1) Fixed `countFailures` in a couple cases where it was getting an incorrect value since it used to be tied to `replace`, eg. when killing executors on a blacklisted node.
2) hard error if you call `sc.killExecutors` with dynamic allocation on, since that's another way the ExecutorAllocationManager and the CoarseGrainedSchedulerBackend would get out of sync.
Added a unit test case which verifies that the calls to ExecutorAllocationClient do not adjust the number of executors.
Author: Imran Rashid <irashid@cloudera.com>
Closes#20604 from squito/SPARK-23365.
## What changes were proposed in this pull request?
If spark is run with "spark.authenticate=true", then it will fail to start in local mode.
This PR generates secret in local mode when authentication on.
## How was this patch tested?
Modified existing unit test.
Manually started spark-shell.
Author: Gabor Somogyi <gabor.g.somogyi@gmail.com>
Closes#20652 from gaborgsomogyi/SPARK-23476.
## What changes were proposed in this pull request?
SPARK-20648 introduced the status `SKIPPED` for the stages. On the UI, previously, skipped stages were shown as `PENDING`; after this change, they are not shown on the UI.
The PR introduce a new section in order to show also `SKIPPED` stages in a proper table.
## How was this patch tested?
manual tests
Author: Marco Gaido <marcogaido91@gmail.com>
Closes#20651 from mgaido91/SPARK-23475.
## What changes were proposed in this pull request?
The root cause of missing completed stages is because `cleanupStages` will never remove skipped stages.
This PR changes the logic to always remove skipped stage first. This is safe since the job itself contains enough information to render skipped stages in the UI.
## How was this patch tested?
The new unit tests.
Author: Shixiong Zhu <zsxwing@gmail.com>
Closes#20656 from zsxwing/SPARK-23475.
## What changes were proposed in this pull request?
The issue here is `AppStatusStore.lastStageAttempt` will return the next available stage in the store when a stage doesn't exist.
This PR adds `last(stageId)` to ensure it returns a correct `StageData`
## How was this patch tested?
The new unit test.
Author: Shixiong Zhu <zsxwing@gmail.com>
Closes#20654 from zsxwing/SPARK-23481.
The secret is used as a string in many parts of the code, so it has
to be turned into a hex string to avoid issues such as the random
byte sequence not containing a valid UTF8 sequence.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20643 from vanzin/SPARK-23468.
This is much faster than finding out what the last attempt is, and the
data should be the same.
There's room for improvement in this page (like only loading data for
the jobs being shown, instead of loading all available jobs and sorting
them), but this should bring performance on par with the 2.2 version.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20644 from vanzin/SPARK-23470.
## What changes were proposed in this pull request?
Print more helpful message when daemon module's stdout is empty or contains a bad port number.
## How was this patch tested?
Manually recreated the environmental issues that caused the mysterious exceptions at one site. Tested that the expected messages are logged.
Also, ran all scala unit tests.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Bruce Robbins <bersprockets@gmail.com>
Closes#20424 from bersprockets/SPARK-23240_prop2.
## What changes were proposed in this pull request?
Fixing exception got at sorting tasks by Host / Executor ID:
```
java.lang.IllegalArgumentException: Invalid sort column: Host
at org.apache.spark.ui.jobs.ApiHelper$.indexName(StagePage.scala:1017)
at org.apache.spark.ui.jobs.TaskDataSource.sliceData(StagePage.scala:694)
at org.apache.spark.ui.PagedDataSource.pageData(PagedTable.scala:61)
at org.apache.spark.ui.PagedTable$class.table(PagedTable.scala:96)
at org.apache.spark.ui.jobs.TaskPagedTable.table(StagePage.scala:708)
at org.apache.spark.ui.jobs.StagePage.liftedTree1$1(StagePage.scala:293)
at org.apache.spark.ui.jobs.StagePage.render(StagePage.scala:282)
at org.apache.spark.ui.WebUI$$anonfun$2.apply(WebUI.scala:82)
at org.apache.spark.ui.WebUI$$anonfun$2.apply(WebUI.scala:82)
at org.apache.spark.ui.JettyUtils$$anon$3.doGet(JettyUtils.scala:90)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:687)
at javax.servlet.http.HttpServlet.service(HttpServlet.java:790)
at org.spark_project.jetty.servlet.ServletHolder.handle(ServletHolder.java:848)
at org.spark_project.jetty.servlet.ServletHandler.doHandle(ServletHandler.java:584)
```
Moreover some refactoring to avoid similar problems by introducing constants for each header name and reusing them at the identification of the corresponding sorting index.
## How was this patch tested?
Manually:
Author: “attilapiros” <piros.attila.zsolt@gmail.com>
Closes#20601 from attilapiros/SPARK-23413.
…D_CLASSES set to 1
## What changes were proposed in this pull request?
YarnShuffleIntegrationSuite fails when SPARK_PREPEND_CLASSES set to 1.
Normally mllib built before yarn module. When SPARK_PREPEND_CLASSES used mllib classes are on yarn test classpath.
Before 2.3 that did not cause issues. But 2.3 has SPARK-22450, which registered some mllib classes with the kryo serializer. Now it dies with the following error:
`
18/02/13 07:33:29 INFO SparkContext: Starting job: collect at YarnShuffleIntegrationSuite.scala:143
Exception in thread "dag-scheduler-event-loop" java.lang.NoClassDefFoundError: breeze/linalg/DenseMatrix
`
In this PR NoClassDefFoundError caught only in case of testing and then do nothing.
## How was this patch tested?
Automated: Pass the Jenkins.
Author: Gabor Somogyi <gabor.g.somogyi@gmail.com>
Closes#20608 from gaborgsomogyi/SPARK-23422.
## What changes were proposed in this pull request?
`ReadAheadInputStream` was introduced in https://github.com/apache/spark/pull/18317/ to optimize reading spill files from disk.
However, from the profiles it seems that the hot path of reading small amounts of data (like readInt) is inefficient - it involves taking locks, and multiple checks.
Optimize locking: Lock is not needed when simply accessing the active buffer. Only lock when needing to swap buffers or trigger async reading, or get information about the async state.
Optimize short-path single byte reads, that are used e.g. by Java library DataInputStream.readInt.
The asyncReader used to call "read" only once on the underlying stream, that never filled the underlying buffer when it was wrapping an LZ4BlockInputStream. If the buffer was returned unfilled, that would trigger the async reader to be triggered to fill the read ahead buffer on each call, because the reader would see that the active buffer is below the refill threshold all the time.
However, filling the full buffer all the time could introduce increased latency, so also add an `AtomicBoolean` flag for the async reader to return earlier if there is a reader waiting for data.
Remove `readAheadThresholdInBytes` and instead immediately trigger async read when switching the buffers. It allows to simplify code paths, especially the hot one that then only has to check if there is available data in the active buffer, without worrying if it needs to retrigger async read. It seems to have positive effect on perf.
## How was this patch tested?
It was noticed as a regression in some workloads after upgrading to Spark 2.3.
It was particularly visible on TPCDS Q95 running on instances with fast disk (i3 AWS instances).
Running with profiling:
* Spark 2.2 - 5.2-5.3 minutes 9.5% in LZ4BlockInputStream.read
* Spark 2.3 - 6.4-6.6 minutes 31.1% in ReadAheadInputStream.read
* Spark 2.3 + fix - 5.3-5.4 minutes 13.3% in ReadAheadInputStream.read - very slightly slower, practically within noise.
We didn't see other regressions, and many workloads in general seem to be faster with Spark 2.3 (not investigated if thanks to async readed, or unrelated).
Author: Juliusz Sompolski <julek@databricks.com>
Closes#20555 from juliuszsompolski/SPARK-23366.
## What changes were proposed in this pull request?
When we run concurrent jobs using the same rdd which is marked to do checkpoint. If one job has finished running the job, and start the process of RDD.doCheckpoint, while another job is submitted, then submitStage and submitMissingTasks will be called. In [submitMissingTasks](https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala#L961), will serialize taskBinaryBytes and calculate task partitions which are both affected by the status of checkpoint, if the former is calculated before doCheckpoint finished, while the latter is calculated after doCheckpoint finished, when run task, rdd.compute will be called, for some rdds with particular partition type such as [UnionRDD](https://github.com/apache/spark/blob/master/core/src/main/scala/org/apache/spark/rdd/UnionRDD.scala) who will do partition type cast, will get a ClassCastException because the part params is actually a CheckpointRDDPartition.
This error occurs because rdd.doCheckpoint occurs in the same thread that called sc.runJob, while the task serialization occurs in the DAGSchedulers event loop.
## How was this patch tested?
the exist uts and also add a test case in DAGScheduerSuite to show the exception case.
Author: huangtengfei <huangtengfei@huangtengfeideMacBook-Pro.local>
Closes#20244 from ivoson/branch-taskpart-mistype.
## What changes were proposed in this pull request?
The purpose of this PR to reflect the stage level blacklisting on the executor tab for the currently active stages.
After this change in the executor tab at the Status column one of the following label will be:
- "Blacklisted" when the executor is blacklisted application level (old flag)
- "Dead" when the executor is not Blacklisted and not Active
- "Blacklisted in Stages: [...]" when the executor is Active but the there are active blacklisted stages for the executor. Within the [] coma separated active stageIDs are listed.
- "Active" when the executor is Active and there is no active blacklisted stages for the executor
## How was this patch tested?
Both with unit tests and manually.
#### Manual test
Spark was started as:
```bash
bin/spark-shell --master "local-cluster[2,1,1024]" --conf "spark.blacklist.enabled=true" --conf "spark.blacklist.stage.maxFailedTasksPerExecutor=1" --conf "spark.blacklist.application.maxFailedTasksPerExecutor=10"
```
And the job was:
```scala
import org.apache.spark.SparkEnv
val pairs = sc.parallelize(1 to 10000, 10).map { x =>
if (SparkEnv.get.executorId.toInt == 0) throw new RuntimeException("Bad executor")
else {
Thread.sleep(10)
(x % 10, x)
}
}
val all = pairs.cogroup(pairs)
all.collect()
```
UI screenshots about the running:
- One executor is blacklisted in the two stages:
- One stage completes the other one is still running:
- Both stages are completed:
### Unit tests
In AppStatusListenerSuite.scala both the node blacklisting for a stage and the executor blacklisting for stage are tested.
Author: “attilapiros” <piros.attila.zsolt@gmail.com>
Closes#20408 from attilapiros/SPARK-23189.
## What changes were proposed in this pull request?
In this PR StorageStatus is made to private and simplified a bit moreover SparkContext.getExecutorStorageStatus method is removed. The reason of keeping StorageStatus is that it is usage from SparkContext.getRDDStorageInfo.
Instead of the method SparkContext.getExecutorStorageStatus executor infos are extended with additional memory metrics such as usedOnHeapStorageMemory, usedOffHeapStorageMemory, totalOnHeapStorageMemory, totalOffHeapStorageMemory.
## How was this patch tested?
By running existing unit tests.
Author: “attilapiros” <piros.attila.zsolt@gmail.com>
Author: Attila Zsolt Piros <2017933+attilapiros@users.noreply.github.com>
Closes#20546 from attilapiros/SPARK-20659.
## What changes were proposed in this pull request?
In the `getBlockData`,`blockId.reduceId` is the `Int` type, when it is greater than 2^28, `blockId.reduceId*8` will overflow
In the `decompress0`, `len` and `unitSize` are Int type, so `len * unitSize` may lead to overflow
## How was this patch tested?
N/A
Author: liuxian <liu.xian3@zte.com.cn>
Closes#20581 from 10110346/overflow2.
## What changes were proposed in this pull request?
Currently, we use SBT and MAVN to spark unit test, are affected by the parameters of `spark.testing`. However, when using the IDE test tool, `spark.testing` support is not very good, sometimes need to be manually added to the beforeEach. example: HiveSparkSubmitSuite RPackageUtilsSuite SparkSubmitSuite. The PR unified `spark.testing` parameter extraction to SparkFunSuite, support IDE test tool, and the test code is more compact.
## How was this patch tested?
the existed test cases.
Author: caoxuewen <cao.xuewen@zte.com.cn>
Closes#20582 from heary-cao/sparktesting.
## What changes were proposed in this pull request?
In the `checkIndexAndDataFile`,the `blocks` is the ` Int` type, when it is greater than 2^28, `blocks*8` will overflow, and this will result in an error result.
In fact, `blocks` is actually the number of partitions.
## How was this patch tested?
Manual test
Author: liuxian <liu.xian3@zte.com.cn>
Closes#20544 from 10110346/overflow.
## What changes were proposed in this pull request?
When `DebugFilesystem` closes opened stream, if any exception occurs, we still need to remove the open stream record from `DebugFilesystem`. Otherwise, it goes to report leaked filesystem connection.
## How was this patch tested?
Existing tests.
Author: Liang-Chi Hsieh <viirya@gmail.com>
Closes#20524 from viirya/SPARK-23345.
## What changes were proposed in this pull request?
This is a follow-up of #20492 which broke lint-java checks.
This pr fixes the lint-java issues.
```
[ERROR] src/main/java/org/apache/spark/util/collection/unsafe/sort/UnsafeSorterSpillReader.java:[79] (sizes) LineLength: Line is longer than 100 characters (found 114).
```
## How was this patch tested?
Checked manually in my local environment.
Author: Takuya UESHIN <ueshin@databricks.com>
Closes#20514 from ueshin/issues/SPARK-23310/fup1.
## What changes were proposed in this pull request?
When a task is still running, metrics like executorRunTime are not available. Then `schedulerDelay` will be almost the same as `duration` and that's confusing.
This PR makes `schedulerDelay` return 0 when the task is running which is the same behavior as 2.2.
## How was this patch tested?
`AppStatusUtilsSuite.schedulerDelay`
Author: Shixiong Zhu <zsxwing@gmail.com>
Closes#20493 from zsxwing/SPARK-23326.
## What changes were proposed in this pull request?
Sort jobs/stages/tasks/queries with the completed timestamp before cleaning up them to make the behavior consistent with 2.2.
## How was this patch tested?
- Jenkins.
- Manually ran the following codes and checked the UI for jobs/stages/tasks/queries.
```
spark.ui.retainedJobs 10
spark.ui.retainedStages 10
spark.sql.ui.retainedExecutions 10
spark.ui.retainedTasks 10
```
```
new Thread() {
override def run() {
spark.range(1, 2).foreach { i =>
Thread.sleep(10000)
}
}
}.start()
Thread.sleep(5000)
for (_ <- 1 to 20) {
new Thread() {
override def run() {
spark.range(1, 2).foreach { i =>
}
}
}.start()
}
Thread.sleep(15000)
spark.range(1, 2).foreach { i =>
}
sc.makeRDD(1 to 100, 100).foreach { i =>
}
```
Author: Shixiong Zhu <zsxwing@gmail.com>
Closes#20481 from zsxwing/SPARK-23307.
## What changes were proposed in this pull request?
Shuffle Index temporay file is used for atomic creating shuffle index file, it is not needed when the index file already exists after another attempts of same task had it done.
## How was this patch tested?
exitsting ut
cc squito
Author: Kent Yao <yaooqinn@hotmail.com>
Closes#20422 from yaooqinn/SPARK-23253.
## What changes were proposed in this pull request?
here is race condition in TaskMemoryManger, which may cause OOM.
The memory released may be taken by another task because there is a gap between releaseMemory and acquireMemory, e.g., UnifiedMemoryManager, causing the OOM. if the current is the only one that can perform spill. It can happen to BytesToBytesMap, as it only spill required bytes.
Loop on current consumer if it still has memory to release.
## How was this patch tested?
The race contention is hard to reproduce, but the current logic seems causing the issue.
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: Zhan Zhang <zhanzhang@fb.com>
Closes#20480 from zhzhan/oom.
First the bad news: there's an unfixable race in the launcher code.
(By unfixable I mean it would take a lot more effort than this change
to fix it.) The good news is that it should only affect super short
lived applications, such as the one run by the flaky test, so it's
possible to work around it in our test.
The fix also uncovered an issue with the recently added "closeAndWait()"
method; closing the connection would still possibly cause data loss,
so this change waits a while for the connection to finish itself, and
closes the socket if that times out. The existing connection timeout
is reused so that if desired it's possible to control how long to wait.
As part of that I also restored the old behavior that disconnect() would
force a disconnection from the child app; the "wait for data to arrive"
approach is only taken when disposing of the handle.
I tested this by inserting a bunch of sleeps in the test and the socket
handling code in the launcher library; with those I was able to reproduce
the error from the jenkins jobs. With the changes, even with all the
sleeps still in place, all tests pass.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20462 from vanzin/SPARK-23020.
## What changes were proposed in this pull request?
`channel.write(buf)` may not write the whole buffer since the underlying channel is a FileChannel, we should retry until the whole buffer is written.
## How was this patch tested?
Jenkins
Author: Shixiong Zhu <zsxwing@gmail.com>
Closes#20461 from zsxwing/SPARK-23289.
## What changes were proposed in this pull request?
Rename the public APIs and names of pandas udfs.
- `PANDAS SCALAR UDF` -> `SCALAR PANDAS UDF`
- `PANDAS GROUP MAP UDF` -> `GROUPED MAP PANDAS UDF`
- `PANDAS GROUP AGG UDF` -> `GROUPED AGG PANDAS UDF`
## How was this patch tested?
The existing tests
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20428 from gatorsmile/renamePandasUDFs.
## What changes were proposed in this pull request?
History server not showing incomplete/running applications when spark.history.ui.maxApplications property is set to a value that is smaller than the total number of applications.
## How was this patch tested?
Verified manually against master and 2.2.2 branch.
Author: Paul Mackles <pmackles@adobe.com>
Closes#20335 from pmackles/SPARK-23088.
The JVM seems to be doing early binding of classes that the Hive provider
depends on, causing an error to be thrown before it was caught by the code
in the class.
The fix wraps the creation of the provider in a try..catch so that
the provider can be ignored when dependencies are missing.
Added a unit test (which fails without the fix), and also tested
that getting tokens still works in a real cluster.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20399 from vanzin/SPARK-23209.
## What changes were proposed in this pull request?
Currently shuffle repartition uses RoundRobinPartitioning, the generated result is nondeterministic since the sequence of input rows are not determined.
The bug can be triggered when there is a repartition call following a shuffle (which would lead to non-deterministic row ordering), as the pattern shows below:
upstream stage -> repartition stage -> result stage
(-> indicate a shuffle)
When one of the executors process goes down, some tasks on the repartition stage will be retried and generate inconsistent ordering, and some tasks of the result stage will be retried generating different data.
The following code returns 931532, instead of 1000000:
```
import scala.sys.process._
import org.apache.spark.TaskContext
val res = spark.range(0, 1000 * 1000, 1).repartition(200).map { x =>
x
}.repartition(200).map { x =>
if (TaskContext.get.attemptNumber == 0 && TaskContext.get.partitionId < 2) {
throw new Exception("pkill -f java".!!)
}
x
}
res.distinct().count()
```
In this PR, we propose a most straight-forward way to fix this problem by performing a local sort before partitioning, after we make the input row ordering deterministic, the function from rows to partitions is fully deterministic too.
The downside of the approach is that with extra local sort inserted, the performance of repartition() will go down, so we add a new config named `spark.sql.execution.sortBeforeRepartition` to control whether this patch is applied. The patch is default enabled to be safe-by-default, but user may choose to manually turn it off to avoid performance regression.
This patch also changes the output rows ordering of repartition(), that leads to a bunch of test cases failure because they are comparing the results directly.
## How was this patch tested?
Add unit test in ExchangeSuite.
With this patch(and `spark.sql.execution.sortBeforeRepartition` set to true), the following query returns 1000000:
```
import scala.sys.process._
import org.apache.spark.TaskContext
spark.conf.set("spark.sql.execution.sortBeforeRepartition", "true")
val res = spark.range(0, 1000 * 1000, 1).repartition(200).map { x =>
x
}.repartition(200).map { x =>
if (TaskContext.get.attemptNumber == 0 && TaskContext.get.partitionId < 2) {
throw new Exception("pkill -f java".!!)
}
x
}
res.distinct().count()
res7: Long = 1000000
```
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20393 from jiangxb1987/shuffle-repartition.
## What changes were proposed in this pull request?
The code logic between `MemoryStore.putIteratorAsValues` and `Memory.putIteratorAsBytes` are almost same, so we should reduce the duplicate code between them.
## How was this patch tested?
Existing UT.
Author: Xianyang Liu <xianyang.liu@intel.com>
Closes#19285 from ConeyLiu/rmemorystore.
## What changes were proposed in this pull request?
Currently,the deserializeStream in ExternalAppendOnlyMap#DiskMapIterator init when DiskMapIterator instance created.This will cause memory use overhead when ExternalAppendOnlyMap spill too much times.
We can avoid this by making deserializeStream init when it is used the first time.
This patch make deserializeStream init lazily.
## How was this patch tested?
Exist tests
Author: zhoukang <zhoukang199191@gmail.com>
Closes#20292 from caneGuy/zhoukang/lay-diskmapiterator.
## What changes were proposed in this pull request?
[Ticket](https://issues.apache.org/jira/browse/SPARK-22297)
- one of the tests seems to produce unreliable results due to execution speed variability
Since the original test was trying to connect to the test server with `40 ms` timeout, and the test server replied after `50 ms`, the error might be produced under the following conditions:
- it might occur that the test server replies correctly after `50 ms`
- but the client does only receive the timeout after `51 ms`s
- this might happen if the executor has to schedule a big number of threads, and decides to delay the thread/actor that is responsible to watch the timeout, because of high CPU load
- running an entire test suite usually produces high loads on the CPU executing the tests
## How was this patch tested?
The test's check cases remain the same and the set-up emulates the previous version's.
Author: Mark Petruska <petruska.mark@gmail.com>
Closes#19671 from mpetruska/SPARK-22297.
## What changes were proposed in this pull request?
This is a follow-up of #20297 which broke lint-java checks.
This pr fixes the lint-java issues.
```
[ERROR] src/test/java/org/apache/spark/launcher/BaseSuite.java:[21,8] (imports) UnusedImports: Unused import - java.util.concurrent.TimeUnit.
[ERROR] src/test/java/org/apache/spark/launcher/SparkLauncherSuite.java:[27,8] (imports) UnusedImports: Unused import - java.util.concurrent.TimeUnit.
```
## How was this patch tested?
Checked manually in my local environment.
Author: Takuya UESHIN <ueshin@databricks.com>
Closes#20376 from ueshin/issues/SPARK-23020/fup1.
## What changes were proposed in this pull request?
In this PR stage blacklisting is propagated to UI by introducing a new Spark listener event (SparkListenerExecutorBlacklistedForStage) which indicates the executor is blacklisted for a stage. Either because of the number of failures are exceeded a limit given for an executor (spark.blacklist.stage.maxFailedTasksPerExecutor) or because of the whole node is blacklisted for a stage (spark.blacklist.stage.maxFailedExecutorsPerNode). In case of the node is blacklisting all executors will listed as blacklisted for the stage.
Blacklisting state for a selected stage can be seen "Aggregated Metrics by Executor" table's blacklisting column, where after this change three possible labels could be found:
- "for application": when the executor is blacklisted for the application (see the configuration spark.blacklist.application.maxFailedTasksPerExecutor for details)
- "for stage": when the executor is **only** blacklisted for the stage
- "false" : when the executor is not blacklisted at all
## How was this patch tested?
It is tested both manually and with unit tests.
#### Unit tests
- HistoryServerSuite
- TaskSetBlacklistSuite
- AppStatusListenerSuite
#### Manual test for executor blacklisting
Running Spark as a local cluster:
```
$ bin/spark-shell --master "local-cluster[2,1,1024]" --conf "spark.blacklist.enabled=true" --conf "spark.blacklist.stage.maxFailedTasksPerExecutor=1" --conf "spark.blacklist.application.maxFailedTasksPerExecutor=10" --conf "spark.eventLog.enabled=true"
```
Executing:
``` scala
import org.apache.spark.SparkEnv
sc.parallelize(1 to 10, 10).map { x =>
if (SparkEnv.get.executorId == "0") throw new RuntimeException("Bad executor")
else (x % 3, x)
}.reduceByKey((a, b) => a + b).collect()
```
To see result check the "Aggregated Metrics by Executor" section at the bottom of picture:
#### Manual test for node blacklisting
Running Spark as on a cluster:
``` bash
./bin/spark-shell --master yarn --deploy-mode client --executor-memory=2G --num-executors=8 --conf "spark.blacklist.enabled=true" --conf "spark.blacklist.stage.maxFailedTasksPerExecutor=1" --conf "spark.blacklist.stage.maxFailedExecutorsPerNode=1" --conf "spark.blacklist.application.maxFailedTasksPerExecutor=10" --conf "spark.eventLog.enabled=true"
```
And the job was:
``` scala
import org.apache.spark.SparkEnv
sc.parallelize(1 to 10000, 10).map { x =>
if (SparkEnv.get.executorId.toInt >= 4) throw new RuntimeException("Bad executor")
else (x % 3, x)
}.reduceByKey((a, b) => a + b).collect()
```
The result is:
Here you can see apiros3.gce.test.com was node blacklisted for the stage because of failures on executor 4 and 5. As expected executor 3 is also blacklisted even it has no failures itself but sharing the node with 4 and 5.
Author: “attilapiros” <piros.attila.zsolt@gmail.com>
Author: Attila Zsolt Piros <2017933+attilapiros@users.noreply.github.com>
Closes#20203 from attilapiros/SPARK-22577.
## What changes were proposed in this pull request?
#20002 purposed a way to safe check the default partitioner, however, if `spark.default.parallelism` is set, the defaultParallelism still could be smaller than the proper number of partitions for upstreams RDDs. This PR tries to extend the approach to address the condition when `spark.default.parallelism` is set.
The requirements where the PR helps with are :
- Max partitioner is not eligible since it is atleast an order smaller, and
- User has explicitly set 'spark.default.parallelism', and
- Value of 'spark.default.parallelism' is lower than max partitioner
- Since max partitioner was discarded due to being at least an order smaller, default parallelism is worse - even though user specified.
Under the rest cases, the changes should be no-op.
## How was this patch tested?
Add corresponding test cases in `PairRDDFunctionsSuite` and `PartitioningSuite`.
Author: Xingbo Jiang <xingbo.jiang@databricks.com>
Closes#20091 from jiangxb1987/partitioner.
## What changes were proposed in this pull request?
Add support for using pandas UDFs with groupby().agg().
This PR introduces a new type of pandas UDF - group aggregate pandas UDF. This type of UDF defines a transformation of multiple pandas Series -> a scalar value. Group aggregate pandas UDFs can be used with groupby().agg(). Note group aggregate pandas UDF doesn't support partial aggregation, i.e., a full shuffle is required.
This PR doesn't support group aggregate pandas UDFs that return ArrayType, StructType or MapType. Support for these types is left for future PR.
## How was this patch tested?
GroupbyAggPandasUDFTests
Author: Li Jin <ice.xelloss@gmail.com>
Closes#19872 from icexelloss/SPARK-22274-groupby-agg.
## What changes were proposed in this pull request?
Typo fixes
## How was this patch tested?
Local build / Doc-only changes
Author: Jacek Laskowski <jacek@japila.pl>
Closes#20344 from jaceklaskowski/typo-fixes.
## What changes were proposed in this pull request?
The allJobs and the job pages attempt to use stage attempt and DAG visualization from the store, but for long running jobs they are not guaranteed to be retained, leading to exceptions when these pages are rendered.
To fix it `store.lastStageAttempt(stageId)` and `store.operationGraphForJob(jobId)` are wrapped in `store.asOption` and default values are used if the info is missing.
## How was this patch tested?
Manual testing of the UI, also using the test command reported in SPARK-23121:
./bin/spark-submit --class org.apache.spark.examples.streaming.HdfsWordCount ./examples/jars/spark-examples_2.11-2.4.0-SNAPSHOT.jar /spark
Closes#20287
Author: Sandor Murakozi <smurakozi@gmail.com>
Closes#20330 from smurakozi/SPARK-23121.
## What changes were proposed in this pull request?
ClosureCleaner moved from warning to debug
## How was this patch tested?
Existing tests
Author: Rekha Joshi <rekhajoshm@gmail.com>
Author: rjoshi2 <rekhajoshm@gmail.com>
Closes#20337 from rekhajoshm/SPARK-11630-1.
The race in the code is because the handle might update
its state to the wrong state if the connection handling
thread is still processing incoming data; so the handle
needs to wait for the connection to finish up before
checking the final state.
The race in the test is because when waiting for a handle
to reach a final state, the waitFor() method needs to wait
until all handle state is updated (which also includes
waiting for the connection thread above to finish).
Otherwise, waitFor() may return too early, which would cause
a bunch of different races (like the listener not being yet
notified of the state change, or being in the middle of
being notified, or the handle not being properly disposed
and causing postChecks() to assert).
On top of that I found, by code inspection, a couple of
potential races that could make a handle end up in the
wrong state when being killed.
The original version of this fix introduced the flipped
version of the first race described above; the connection
closing might override the handle state before the
handle might have a chance to do cleanup. The fix there
is to only dispose of the handle from the connection
when there is an error, and let the handle dispose
itself in the normal case.
The fix also caused a bug in YarnClusterSuite to be surfaced;
the code was checking for a file in the classpath that was
not expected to be there in client mode. Because of the above
issues, the error was not propagating correctly and the (buggy)
test was incorrectly passing.
Tested by running the existing unit tests a lot (and not
seeing the errors I was seeing before).
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20297 from vanzin/SPARK-23020.
## What changes were proposed in this pull request?
The clean up logic on the worker perviously determined the liveness of a
particular applicaiton based on whether or not it had running executors.
This would fail in the case that a directory was made for a driver
running in cluster mode if that driver had no running executors on the
same machine. To preserve driver directories we consider both executors
and running drivers when checking directory liveness.
## How was this patch tested?
Manually started up two node cluster with a single core on each node. Turned on worker directory cleanup and set the interval to 1 second and liveness to one second. Without the patch the driver directory is removed immediately after the app is launched. With the patch it is not
### Without Patch
```
INFO 2018-01-05 23:48:24,693 Logging.scala:54 - Asked to launch driver driver-20180105234824-0000
INFO 2018-01-05 23:48:25,293 Logging.scala:54 - Changing view acls to: cassandra
INFO 2018-01-05 23:48:25,293 Logging.scala:54 - Changing modify acls to: cassandra
INFO 2018-01-05 23:48:25,294 Logging.scala:54 - Changing view acls groups to:
INFO 2018-01-05 23:48:25,294 Logging.scala:54 - Changing modify acls groups to:
INFO 2018-01-05 23:48:25,294 Logging.scala:54 - SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(cassandra); groups with view permissions: Set(); users with modify permissions: Set(cassandra); groups with modify permissions: Set()
INFO 2018-01-05 23:48:25,330 Logging.scala:54 - Copying user jar file:/home/automaton/writeRead-0.1.jar to /var/lib/spark/worker/driver-20180105234824-0000/writeRead-0.1.jar
INFO 2018-01-05 23:48:25,332 Logging.scala:54 - Copying /home/automaton/writeRead-0.1.jar to /var/lib/spark/worker/driver-20180105234824-0000/writeRead-0.1.jar
INFO 2018-01-05 23:48:25,361 Logging.scala:54 - Launch Command: "/usr/lib/jvm/jdk1.8.0_40//bin/java" ....
****
INFO 2018-01-05 23:48:56,577 Logging.scala:54 - Removing directory: /var/lib/spark/worker/driver-20180105234824-0000 ### << Cleaned up
****
--
One minute passes while app runs (app has 1 minute sleep built in)
--
WARN 2018-01-05 23:49:58,080 ShuffleSecretManager.java:73 - Attempted to unregister application app-20180105234831-0000 when it is not registered
INFO 2018-01-05 23:49:58,081 ExternalShuffleBlockResolver.java:163 - Application app-20180105234831-0000 removed, cleanupLocalDirs = false
INFO 2018-01-05 23:49:58,081 ExternalShuffleBlockResolver.java:163 - Application app-20180105234831-0000 removed, cleanupLocalDirs = false
INFO 2018-01-05 23:49:58,082 ExternalShuffleBlockResolver.java:163 - Application app-20180105234831-0000 removed, cleanupLocalDirs = true
INFO 2018-01-05 23:50:00,999 Logging.scala:54 - Driver driver-20180105234824-0000 exited successfully
```
With Patch
```
INFO 2018-01-08 23:19:54,603 Logging.scala:54 - Asked to launch driver driver-20180108231954-0002
INFO 2018-01-08 23:19:54,975 Logging.scala:54 - Changing view acls to: automaton
INFO 2018-01-08 23:19:54,976 Logging.scala:54 - Changing modify acls to: automaton
INFO 2018-01-08 23:19:54,976 Logging.scala:54 - Changing view acls groups to:
INFO 2018-01-08 23:19:54,976 Logging.scala:54 - Changing modify acls groups to:
INFO 2018-01-08 23:19:54,976 Logging.scala:54 - SecurityManager: authentication disabled; ui acls disabled; users with view permissions: Set(automaton); groups with view permissions: Set(); users with modify permissions: Set(automaton); groups with modify permissions: Set()
INFO 2018-01-08 23:19:55,029 Logging.scala:54 - Copying user jar file:/home/automaton/writeRead-0.1.jar to /var/lib/spark/worker/driver-20180108231954-0002/writeRead-0.1.jar
INFO 2018-01-08 23:19:55,031 Logging.scala:54 - Copying /home/automaton/writeRead-0.1.jar to /var/lib/spark/worker/driver-20180108231954-0002/writeRead-0.1.jar
INFO 2018-01-08 23:19:55,038 Logging.scala:54 - Launch Command: ......
INFO 2018-01-08 23:21:28,674 ShuffleSecretManager.java:69 - Unregistered shuffle secret for application app-20180108232000-0000
INFO 2018-01-08 23:21:28,675 ExternalShuffleBlockResolver.java:163 - Application app-20180108232000-0000 removed, cleanupLocalDirs = false
INFO 2018-01-08 23:21:28,675 ExternalShuffleBlockResolver.java:163 - Application app-20180108232000-0000 removed, cleanupLocalDirs = false
INFO 2018-01-08 23:21:28,681 ExternalShuffleBlockResolver.java:163 - Application app-20180108232000-0000 removed, cleanupLocalDirs = true
INFO 2018-01-08 23:21:31,703 Logging.scala:54 - Driver driver-20180108231954-0002 exited successfully
*****
INFO 2018-01-08 23:21:32,346 Logging.scala:54 - Removing directory: /var/lib/spark/worker/driver-20180108231954-0002 ### < Happening AFTER the Run completes rather than during it
*****
```
Author: Russell Spitzer <Russell.Spitzer@gmail.com>
Closes#20298 from RussellSpitzer/SPARK-22976-master.
The code was sorting "0" as "less than" negative values, which is a little
wrong. Fix is simple, most of the changes are the added test and related
cleanup.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20284 from vanzin/SPARK-23103.
Detect the deletion of event log files from storage, and remove
data about the related application attempt in the SHS.
Also contains code to fix SPARK-21571 based on code by ericvandenbergfb.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20138 from vanzin/SPARK-20664.
## What changes were proposed in this pull request?
Spark ui about the contents of the form need to have hidden and show features, when the table records very much. Because sometimes you do not care about the record of the table, you just want to see the contents of the next table, but you have to scroll the scroll bar for a long time to see the contents of the next table.
Currently we have about 500 workers, but I just wanted to see the logs for the running applications table. I had to scroll through the scroll bars for a long time to see the logs for the running applications table.
In order to ensure functional consistency, I modified the Master Page, Worker Page, Job Page, Stage Page, Task Page, Configuration Page, Storage Page, Pool Page.
fix before:
fix after:
## How was this patch tested?
manual tests
Please review http://spark.apache.org/contributing.html before opening a pull request.
Author: guoxiaolong <guo.xiaolong1@zte.com.cn>
Closes#20216 from guoxiaolongzte/SPARK-23024.
## What changes were proposed in this pull request?
This PR completes the docs, specifying the default units assumed in configuration entries of type size.
This is crucial since unit-less values are accepted and the user might assume the base unit is bytes, which in most cases it is not, leading to hard-to-debug problems.
## How was this patch tested?
This patch updates only documentation only.
Author: Fernando Pereira <fernando.pereira@epfl.ch>
Closes#20269 from ferdonline/docs_units.
## What changes were proposed in this pull request?
Stage's task page table will throw an exception when there's no complete tasks. Furthermore, because the `dataSize` doesn't take running tasks into account, so sometimes UI cannot show the running tasks. Besides table will only be displayed when first task is finished according to the default sortColumn("index").
To reproduce this issue, user could try `sc.parallelize(1 to 20, 20).map { i => Thread.sleep(10000); i }.collect()` or `sc.parallelize(1 to 20, 20).map { i => Thread.sleep((20 - i) * 1000); i }.collect` to reproduce the above issue.
Here propose a solution to fix it. Not sure if it is a right fix, please help to review.
## How was this patch tested?
Manual test.
Author: jerryshao <sshao@hortonworks.com>
Closes#20315 from jerryshao/SPARK-23147.
## What changes were proposed in this pull request?
Temporarily ignoring flaky test `SparkLauncherSuite.testInProcessLauncher` to de-flake the builds. This should be re-enabled when SPARK-23020 is merged.
## How was this patch tested?
N/A (Test Only Change)
Author: Sameer Agarwal <sameerag@apache.org>
Closes#20291 from sameeragarwal/disable-test-2.
## What changes were proposed in this pull request?
Lots of our tests don't properly shutdown everything they create, and end up leaking lots of threads. For example, `TaskSetManagerSuite` doesn't stop the extra `TaskScheduler` and `DAGScheduler` it creates. There are a couple more instances, eg. in `DAGSchedulerSuite`.
This PR adds the possibility to print out the not properly stopped thread list after a test suite executed. The format is the following:
```
===== FINISHED o.a.s.scheduler.DAGSchedulerSuite: 'task end event should have updated accumulators (SPARK-20342)' =====
...
===== Global thread whitelist loaded with name /thread_whitelist from classpath: rpc-client.*, rpc-server.*, shuffle-client.*, shuffle-server.*' =====
ScalaTest-run:
===== THREADS NOT STOPPED PROPERLY =====
ScalaTest-run: dag-scheduler-event-loop
ScalaTest-run: globalEventExecutor-2-5
ScalaTest-run:
===== END OF THREAD DUMP =====
ScalaTest-run:
===== EITHER PUT THREAD NAME INTO THE WHITELIST FILE OR SHUT IT DOWN PROPERLY =====
```
With the help of this leaking threads has been identified in TaskSetManagerSuite. My intention is to hunt down and fix such bugs in later PRs.
## How was this patch tested?
Manual: TaskSetManagerSuite test executed and found out where are the leaking threads.
Automated: Pass the Jenkins.
Author: Gabor Somogyi <gabor.g.somogyi@gmail.com>
Closes#19893 from gaborgsomogyi/SPARK-16139.
The race in the code is because the handle might update
its state to the wrong state if the connection handling
thread is still processing incoming data; so the handle
needs to wait for the connection to finish up before
checking the final state.
The race in the test is because when waiting for a handle
to reach a final state, the waitFor() method needs to wait
until all handle state is updated (which also includes
waiting for the connection thread above to finish).
Otherwise, waitFor() may return too early, which would cause
a bunch of different races (like the listener not being yet
notified of the state change, or being in the middle of
being notified, or the handle not being properly disposed
and causing postChecks() to assert).
On top of that I found, by code inspection, a couple of
potential races that could make a handle end up in the
wrong state when being killed.
Tested by running the existing unit tests a lot (and not
seeing the errors I was seeing before).
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20223 from vanzin/SPARK-23020.
## What changes were proposed in this pull request?
In 2.2, Spark UI displayed the stage description if the job description was not set. This functionality was broken, the GUI has shown no description in this case. In addition, the code uses jobName and
jobDescription instead of stageName and stageDescription when JobTableRowData is created.
In this PR the logic producing values for the job rows was modified to find the latest stage attempt for the job and use that as a fallback if job description was missing.
StageName and stageDescription are also set using values from stage and jobName/description is used only as a fallback.
## How was this patch tested?
Manual testing of the UI, using the code in the bug report.
Author: Sandor Murakozi <smurakozi@gmail.com>
Closes#20251 from smurakozi/SPARK-23051.
## What changes were proposed in this pull request?
We are now forced to use `pyspark/daemon.py` and `pyspark/worker.py` in PySpark.
This doesn't allow a custom modification for it (well, maybe we can still do this in a super hacky way though, for example, setting Python executable that has the custom modification). Because of this, for example, it's sometimes hard to debug what happens inside Python worker processes.
This is actually related with [SPARK-7721](https://issues.apache.org/jira/browse/SPARK-7721) too as somehow Coverage is unable to detect the coverage from `os.fork`. If we have some custom fixes to force the coverage, it works fine.
This is also related with [SPARK-20368](https://issues.apache.org/jira/browse/SPARK-20368). This JIRA describes Sentry support which (roughly) needs some changes within worker side.
With this configuration advanced users will be able to do a lot of pluggable workarounds and we can meet such potential needs in the future.
As an example, let's say if I configure the module `coverage_daemon` and had `coverage_daemon.py` in the python path:
```python
import os
from pyspark import daemon
if "COVERAGE_PROCESS_START" in os.environ:
from pyspark.worker import main
def _cov_wrapped(*args, **kwargs):
import coverage
cov = coverage.coverage(
config_file=os.environ["COVERAGE_PROCESS_START"])
cov.start()
try:
main(*args, **kwargs)
finally:
cov.stop()
cov.save()
daemon.worker_main = _cov_wrapped
if __name__ == '__main__':
daemon.manager()
```
I can track the coverages in worker side too.
More importantly, we can leave the main code intact but allow some workarounds.
## How was this patch tested?
Manually tested.
Author: hyukjinkwon <gurwls223@gmail.com>
Closes#20151 from HyukjinKwon/configuration-daemon-worker.
## What changes were proposed in this pull request?
This pr to make `0` as a valid value for `spark.dynamicAllocation.executorIdleTimeout`.
For details, see the jira description: https://issues.apache.org/jira/browse/SPARK-22870.
## How was this patch tested?
N/A
Author: Yuming Wang <yumwang@ebay.com>
Author: Yuming Wang <wgyumg@gmail.com>
Closes#20080 from wangyum/SPARK-22870.
## What changes were proposed in this pull request?
Spark still use a few years old version 3.2.11. This change is to upgrade json4s to 3.5.3.
Note that this change does not include the Jackson update because the Jackson version referenced in json4s 3.5.3 is 2.8.4, which has a security vulnerability ([see](https://issues.apache.org/jira/browse/SPARK-20433)).
## How was this patch tested?
Existing unit tests and build.
Author: shimamoto <chibochibo@gmail.com>
Closes#20233 from shimamoto/upgrade-json4s.
## What changes were proposed in this pull request?
This patch bumps the master branch version to `2.4.0-SNAPSHOT`.
## How was this patch tested?
N/A
Author: gatorsmile <gatorsmile@gmail.com>
Closes#20222 from gatorsmile/bump24.
When resources happen to be constrained on an executor the first time a broadcast variable is instantiated it is persisted to disk by the BlockManager. Consequently, every subsequent call to TorrentBroadcast::readBroadcastBlock from other instances of that broadcast variable spawns another instance of the underlying value. That is, broadcast variables are spawned once per executor **unless** memory is constrained, in which case every instance of a broadcast variable is provided with a unique copy of the underlying value.
This patch fixes the above by explicitly caching the underlying values using weak references in a ReferenceMap.
Author: ho3rexqj <ho3rexqj@gmail.com>
Closes#20183 from ho3rexqj/fix/cache-broadcast-values.
There are two main changes to speed up rendering of the tasks list
when rendering the stage page.
The first one makes the code only load the tasks being shown in the
current page of the tasks table, and information related to only
those tasks. One side-effect of this change is that the graph that
shows task-related events now only shows events for the tasks in
the current page, instead of the previously hardcoded limit of "events
for the first 1000 tasks". That ends up helping with readability,
though.
To make sorting efficient when using a disk store, the task wrapper
was extended to include many new indices, one for each of the sortable
columns in the UI, and metrics for which quantiles are calculated.
The second changes the way metric quantiles are calculated for stages.
Instead of using the "Distribution" class to process data for all task
metrics, which requires scanning all tasks of a stage, the code now
uses the KVStore "skip()" functionality to only read tasks that contain
interesting information for the quantiles that are desired.
This is still not cheap; because there are many metrics that the UI
and API track, the code needs to scan the index for each metric to
gather the information. Savings come mainly from skipping deserialization
when using the disk store, but the in-memory code also seems to be
faster than before (most probably because of other changes in this
patch).
To make subsequent calls faster, some quantiles are cached in the
status store. This makes UIs much faster after the first time a stage
has been loaded.
With the above changes, a lot of code in the UI layer could be simplified.
Author: Marcelo Vanzin <vanzin@cloudera.com>
Closes#20013 from vanzin/SPARK-20657.
