* Use OverlordClient for all Overlord RPCs.
Continuing the work from #12696, this patch removes HttpIndexingServiceClient
and the IndexingService flavor of DruidLeaderClient completely. All remaining
usages are migrated to OverlordClient.
Supporting changes include:
1) Add a variety of methods to OverlordClient.
2) Update MetadataTaskStorage to skip the complete-task lookup when
the caller requests zero completed tasks. This helps performance of
the "get active tasks" APIs, which don't want to see complete ones.
* Use less forbidden APIs.
* Fixes from CI.
* Add test coverage.
* Two more tests.
* Fix test.
* Updates from CR.
* Remove unthrown exceptions.
* Refactor to improve testability and test coverage.
* Add isNil tests.
* Remove unnecessary "deserialize" methods.
* Add ingest/input/bytes metric and Kafka consumer metrics.
New metrics:
1) ingest/input/bytes. Equivalent to processedBytes in the task reports.
2) kafka/consumer/bytesConsumed: Equivalent to the Kafka consumer
metric "bytes-consumed-total". Only emitted for Kafka tasks.
3) kafka/consumer/recordsConsumed: Equivalent to the Kafka consumer
metric "records-consumed-total". Only emitted for Kafka tasks.
* Fix anchor.
* Fix KafkaConsumerMonitor.
* Interface updates.
* Doc changes.
* Update indexing-service/src/main/java/org/apache/druid/indexing/seekablestream/SeekableStreamIndexTask.java
Co-authored-by: Benedict Jin <asdf2014@apache.org>
---------
Co-authored-by: Benedict Jin <asdf2014@apache.org>
changes:
* new filters that preserve match value typing to better handle filtering different column types
* sql planner uses new filters by default in sql compatible null handling mode
* remove isFilterable from column capabilities
* proper handling of array filtering, add array processor to column processors
* javadoc for sql test filter functions
* range filter support for arrays, tons more tests, fixes
* add dimension selector tests for mixed type roots
* support json equality
* rename semantic index maker thingys to mostly have plural names since they typically make many indexes, e.g. StringValueSetIndex -> StringValueSetIndexes
* add cooler equality index maker, ValueIndexes
* fix missing string utf8 index supplier
* expression array comparator stuff
* More efficient generation of ImmutableWorkerHolder from WorkerHolder.
Taking the work done in #12096 a little further:
1) Applying a similar optimization to WorkerHolder (HttpRemoteTaskRunner).
The original patch only helped with the ZkWorker (RemoteTaskRunner).
2) Improve the ZkWorker version somewhat by avoiding multiple iterations
through the task announcements map.
* Pick better names and use better logic.
* Only runnable tasks.
* Fix test.
* Fix testBlacklistZKWorkers50Percent.
* Claim full support for Java 17.
No production code has changed, except the startup scripts.
Changes:
1) Allow Java 17 without DRUID_SKIP_JAVA_CHECK.
2) Include the full list of opens and exports on both Java 11 and 17.
3) Document that Java 17 is both supported and preferred.
4) Switch some tests from Java 11 to 17 to get better coverage on the
preferred version.
* Doc update.
* Update errorprone.
* Update docker_build_containers.sh.
* Update errorprone in licenses.yaml.
* Add some more run-javas.
* Additional run-javas.
* Update errorprone.
* Suppress new errorprone error.
* Add exports and opens in ForkingTaskRunner for Java 11+.
Test, doc changes.
* Additional errorprone updates.
* Update for errorprone.
* Restore old fomatting in LdapCredentialsValidator.
* Copy bin/ too.
* Fix Java 15, 17 build line in docker_build_containers.sh.
* Update busybox image.
* One more java command.
* Fix interpolation.
* IT commandline refinements.
* Switch to busybox 1.34.1-glibc.
* POM adjustments, build and test one IT on 17.
* Additional debugging.
* Fix silly thing.
* Adjust command line.
* Add exports and opens one more place.
* Additional harmonization of strong encapsulation parameters.
Recently #14532 fixed a problem when maxLazyWorkers == 0 and lazyWorkers
starts out empty. Unfortunately, even after that patch, there remained
a more general version of this problem when maxLazyWorkers == lazyWorkers.size().
This patch fixes it.
I'm not sure if this would actually happen in production, because the
provisioning strategies do try to avoid calling markWorkersLazy until
previously-initiated terminations have finished. Nevertheless, it still
seems like a good thing to fix.
Description:
`TaskQueue.notifyStatus` is often a heavy call as it performs the following operations:
- Update task status in metadata DB
- Update task locks in metadata DB
- Request (synchronously) the task runner to shutdown the completed task
- Clean up in-memory data structures
This method can often be slow and can cause worker sync / task runners to slow down.
Main changes:
- Run task completion callbacks in a separate executor to handle task completion updates
- Add new config `druid.indexer.queue.taskCompleteHandlerNumThreads`
- Add metrics to monitor number of processed and queued items
- There are still other paths that can invoke `notifyStatus`, but those need not be moved to
the new executor as they are synchronous on purpose.
Other changes:
- Add new metrics `task/status/queue/count`, `task/status/handled/count`
- Add `TaskCountStatsProvider.getStats()` which deprecates the other `getXXXTaskCount` methods.
- Use `CoordinatorRunStats` to collect and report metrics. This class has been used as is
for now but will later be renamed and repurposed to use across all Druid services.
The wait doesn't seem to serve a purpose, other than causing delays
when checking isInitialized() for a large number of things that have
not yet been initialized.
If a server is removed during `HttpServerInventoryView.serverInventoryInitialized`,
the initialization gets stuck as this server is never synced. The method eventually times
out (default 250s).
Fix: Mark a server as stopped if it is removed. `serverInventoryInitialized` only waits for
non-stopped servers to sync.
Other changes:
- Add new metrics for better debugging of slow broker/coordinator startup
- `segment/serverview/sync/healthy`: whether the server view is syncing properly with a server
- `segment/serverview/sync/unstableTime`: time for which sync with a server has been unstable
- Clean up logging in `HttpServerInventoryView` and `ChangeRequestHttpSyncer`
- Minor refactor for readability
- Add utility class `Stopwatch`
- Add tests and stubs
The latest topic offsets are polled frequently and used to determine the lag based on the current offsets. However, when the offsets are stale (which can happen due to connection issues commonly), we may see a negative lag .
This PR prevents emission of metrics when the offsets are stale and at least one of the partitions has a negative lag.
* combine string column implementations
changes:
* generic indexed, front-coded, and auto string columns now all share the same column and index supplier implementations
* remove CachingIndexed implementation, which I think is largely no longer needed by the switch of many things to directly using ByteBuffer, avoiding the cost of creating Strings
* remove ColumnConfig.columnCacheSizeBytes since CachingIndexed was the only user
This PR aims to expose a new API called
"@path("/druid/v2/sql/statements/")" which takes the same payload as the current "/druid/v2/sql" endpoint and allows users to fetch results in an async manner.
Introduce DruidException, an exception whose goal in life is to be delivered to a user.
DruidException itself has javadoc on it to describe how it should be used. This commit both introduces the Exception and adjusts some of the places that are generating exceptions to generate DruidException objects instead, as a way to show how the Exception should be used.
This work was a 3rd iteration on top of work that was started by Paul Rogers. I don't know if his name will survive the squash-and-merge, so I'm calling it out here and thanking him for starting on this.
They were not previously loaded because supportsQueries was false.
This patch sets supportsQueries to true, and clarifies in Task
javadocs that supportsQueries can be true for tasks that aren't
directly queryable over HTTP.
changes:
* auto columns no longer participate in generic 'null column' handling, this was a mistake to try to support and caused ingestion failures due to mismatched ColumnFormat, and will be replaced in the future with nested common format constant column functionality (not in this PR)
* fix bugs with auto columns which contain empty objects, empty arrays, or primitive types mixed with either of these empty constructs
* fix bug with bound filter when upper is null equivalent but is strict
Changes
- Add a `DruidException` which contains a user-facing error message, HTTP response code
- Make `EntryExistsException` extend `DruidException`
- If metadata store max_allowed_packet limit is violated while inserting a new task, throw
`DruidException` with response code 400 (bad request) to prevent retries
- Add `SQLMetadataConnector.isRootCausePacketTooBigException` with impl for MySQL
In this PR, we are enhancing KafkaEmitter, to emit metadata about published segments (SegmentMetadataEvent) into a Kafka topic. This segment metadata information that gets published into Kafka, can be used by any other downstream services to query Druid intelligently based on the segments published. The segment metadata gets published into kafka topic in json string format similar to other events.
The sampler API returns a `400 bad request` response if it encounters a `SamplerException`.
Otherwise, it returns a generic `500 Internal server error` response, with the message
"The RuntimeException could not be mapped to a response, re-throwing to the HTTP container".
This commit updates `RecordSupplierInputSource` to handle all types of exceptions instead of just
`InterruptedException`and wrap them in a `SamplerException` so that the actual error is
propagated back to the user.
It was found that several supported tasks / input sources did not have implementations for the methods used by the input source security feature, causing these tasks and input sources to fail when used with this feature. This pr adds the needed missing implementations. Also securing the sampling endpoint with input source security, when enabled.
* Be able to load segments on Peons
This change introduces a new config on WorkerConfig
that indicates how many bytes of each storage
location to use for storage of a task. Said config
is divided up amongst the locations and slots
and then used to set TaskConfig.tmpStorageBytesPerTask
The Peons use their local task dir and
tmpStorageBytesPerTask as their StorageLocations for
the SegmentManager such that they can accept broadcast
segments.
Changes:
- Replace `OverlordHelper` with `OverlordDuty` to align with `CoordinatorDuty`
- Each duty has a `run()` method and defines a `Schedule` with an initial delay and period.
- Update existing duties `TaskLogAutoCleaner` and `DurableStorageCleaner`
- Add utility class `Configs`
- Update log, error messages and javadocs
- Other minor style improvements
In StreamChunkParser#parseWithInputFormat, we call byteEntityReader.read() without handling a potential ParseException, which is thrown during this function call by the delegate AvroStreamReader#intermediateRowIterator.
A ParseException can be thrown if an Avro stream has corrupt data or data that doesn't conform to the schema specified or for other decoding reasons. This exception if uncaught, can cause ingestion to fail.
This PR fixes an issue when using 'auto' encoded LONG typed columns and the 'vectorized' query engine. These columns use a delta based bit-packing mechanism, and errors in the vectorized reader would cause it to incorrectly read column values for some bit sizes (1 through 32 bits). This is a regression caused by #11004, which added the optimized readers to improve performance, so impacts Druid versions 0.22.0+.
While writing the test I finally got sad enough about IndexSpec not having a "builder", so I made one, and switched all the things to use it. Apologies for the noise in this bug fix PR, the only real changes are in VSizeLongSerde, and the tests that have been modified to cover the buggy behavior, VSizeLongSerdeTest and ExpressionVectorSelectorsTest. Everything else is just cleanup of IndexSpec usage.
* Make LoggingEmitter more useful
* Skip code coverage for facade classes
* fix spellcheck
* code review
* fix dependency
* logging.md
* fix checkstyle
* Add back jacoco version to main pom
Hadoop 2 often causes red security scans on Druid distribution because of the dependencies it brings. We want to move away from Hadoop 2 and provide Hadoop 3 distribution available. Switch druid to building with Hadoop 3 by default. Druid will still be compatible with Hadoop 2 and users can build hadoop-2 compatible distribution using hadoop2 profile.
* Compaction: Block input specs not aligned with segmentGranularity.
When input intervals are not aligned with segmentGranularity, data may be
overshadowed if it lies in the space between the input intervals and the
output segmentGranularity.
In MSQ REPLACE, this is a validation error. IMO the same behavior makes
sense for compaction tasks. In case anyone was depending on the ability
to compact nonaligned intervals, a configuration parameter
allowNonAlignedInterval is provided. I don't expect it to be used much.
* Remove unused.
* ITCompactionTaskTest uses non-aligned intervals.
* return task status reported by peon
* Write TaskStatus to file in AbstractTask.cleanUp
* Get TaskStatus from task log
* Fix merge conflicts in AbstractTaskTest
* Add unit tests for TaskLogPusher, TaskLogStreamer, NoopTaskLogs to satisfy code coverage
* Add license headerss
* Fix style
* Remove unknown exception declarations
* Allow for Log4J to be configured for peons but still ensure console logging is enforced
This change will allow for log4j to be configured for peons but require console logging is still
configured for them to ensure peon logs are saved to deep storage.
Also fixed the test ConsoleLoggingEnforcementTest to use a valid appender for the non console
Config as the previous config was incorrect and would never return a logger.
* fix checkstyle
* add warning to logger when it overwrites all loggers to be console
* optimize calls for altering logging config for ConsoleLoggingEnforcementConfigurationFactory
add getName to the druid logger class
* update docs, and error message
* edit docs to be more clear
* fix checkstyle issues
* CI fixes - LoggerTest code coverage and fix spelling issue for logging docs
### Description
This pr fixes a few bugs found with the inputSource security feature.
1. `KillUnusedSegmentsTask` previously had no definition for the `getInputSourceResources`, which caused an unsupportedOperationException to be thrown when this task type was submitted with the inputSource security feature enabled. This task type should not require any input source specific resources, so returning an empty set for this task type now.
2. Fixed a bug where when the input source type security feature is enabled, all of the input source type specific resources used where authenticated against:
`{"resource": {"name": "EXTERNAL", "type": "{INPUT_SOURCE_TYPE}"}, "action": "READ"}`
When they should be instead authenticated against:
`{"resource": {"name": "{INPUT_SOURCE_TYPE}", "type": "EXTERNAL"}, "action": "READ"}`
3. fixed bug where supervisor tasks were not authenticated against the specific input source types used, if input source security feature was enabled.
*
Adds new run time parameter druid.indexer.task.tmpStorageBytesPerTask. This sets a limit for the amount of temporary storage disk space used by tasks. This limit is currently only respected by MSQ tasks.
* Removes query context parameters intermediateSuperSorterStorageMaxLocalBytes and composedIntermediateSuperSorterStorageEnabled. Composed intermediate super sorter (which was enabled by composedIntermediateSuperSorterStorageEnabled) is now enabled automatically if durableShuffleStorage is set to true. intermediateSuperSorterStorageMaxLocalBytes is calculated from the limit set by the run time parameter druid.indexer.task.tmpStorageBytesPerTask.
* Make the tasks run with only a single directory
There was a change that tried to get indexing to run on multiple disks
It made a bunch of changes to how tasks run, effectively hiding the
"safe" directory for tasks to write files into from the task code itself
making it extremely difficult to do anything correctly inside of a task.
This change reverts those changes inside of the tasks and makes it so that
only the task runners are the ones that make decisions about which
mount points should be used for storing task-related files.
It adds the config druid.worker.baseTaskDirs which can be used by the
task runners to know which directories they should schedule tasks inside of.
The TaskConfig remains the authoritative source of configuration for where
and how an individual task should be operating.
changes:
* adds support for boolean inputs to the classic long dimension indexer, which plays nice with LONG being the semi official boolean type in Druid, and even nicer when druid.expressions.useStrictBooleans is set to true, since the sampler when using the new 'auto' schema when 'useSchemaDiscovery' is specified on the dimensions spec will call the type out as LONG
* fix bugs with sampler response and new schema discovery stuff incorrectly using classic 'json' type for the logical schema instead of the new 'auto' type
Fixes#13837.
### Description
This change allows for input source type security in the native task layer.
To enable this feature, the user must set the following property to true:
`druid.auth.enableInputSourceSecurity=true`
The default value for this property is false, which will continue the existing functionality of needing authorization to write to the respective datasource.
When this config is enabled, the users will be required to be authorized for the following resource action, in addition to write permission on the respective datasource.
`new ResourceAction(new Resource(ResourceType.EXTERNAL, {INPUT_SOURCE_TYPE}, Action.READ`
where `{INPUT_SOURCE_TYPE}` is the type of the input source being used;, http, inline, s3, etc..
Only tasks that provide a non-default implementation of the `getInputSourceResources` method can be submitted when config `druid.auth.enableInputSourceSecurity=true` is set. Otherwise, a 400 error will be thrown.
changes:
* introduce ColumnFormat to separate physical storage format from logical type. ColumnFormat is now used instead of ColumnCapabilities to get column handlers for segment creation
* introduce new 'auto' type indexer and merger which produces a new common nested format of columns, which is the next logical iteration of the nested column stuff. Essentially this is an automatic type column indexer that produces the most appropriate column for the given inputs, making either STRING, ARRAY<STRING>, LONG, ARRAY<LONG>, DOUBLE, ARRAY<DOUBLE>, or COMPLEX<json>.
* revert NestedDataColumnIndexer, NestedDataColumnMerger, NestedDataColumnSerializer to their version pre #13803 behavior (v4) for backwards compatibility
* fix a bug in RoaringBitmapSerdeFactory if anything actually ever wrote out an empty bitmap using toBytes and then later tried to read it (the nerve!)
Changes:
- Set `useRoundRobinSegmentAssignment` in coordinator dynamic config to `true` by default.
- Set `batchSegmentAllocation` in `TaskLockConfig` (used in Overlord runtime properties) to `true` by default.
* Lower default maxRowsInMemory for realtime ingestion.
The thinking here is that for best ingestion throughput, we want
intermediate persists to be as big as possible without using up all
available memory. So, we rely mainly on maxBytesInMemory. The default
maxRowsInMemory (1 million) is really just a safety: in case we have
a large number of very small rows, we don't want to get overwhelmed
by per-row overheads.
However, maximum ingestion throughput isn't necessarily the primary
goal for realtime ingestion. Query performance is also important. And
because query performance is not as good on the in-memory dataset, it's
helpful to keep it from growing too large. 150k seems like a reasonable
balance here. It means that for a typical 5 million row segment, we
won't trigger more than 33 persists due to this limit, which is a
reasonable number of persists.
* Update tests.
* Update server/src/main/java/org/apache/druid/segment/indexing/RealtimeTuningConfig.java
Co-authored-by: Kashif Faraz <kashif.faraz@gmail.com>
* Fix test.
* Fix link.
---------
Co-authored-by: Kashif Faraz <kashif.faraz@gmail.com>
* As a follow up to #13893, this PR improves the comments added along with examples for the code, as well as adds handling for an edge case where the generated tombstone boundaries were overshooting the bounds of MIN_TIME (or MAX_TIME).
* Improve memory efficiency of WrappedRoaringBitmap.
Two changes:
1) Use an int[] for sizes 4 or below.
2) Remove the boolean compressRunOnSerialization. Doesn't save much
space, but it does save a little, and it isn't adding a ton of value
to have it be configurable. It was originally configurable in case
anything broke when enabling it, but it's been a while and nothing
has broken.
* Slight adjustment.
* Adjust for inspection.
* Updates.
* Update snaps.
* Update test.
* Adjust test.
* Fix snaps.
* Use TaskConfig to get task dir in KubernetesTaskRunner
* Use the first path specified in baseTaskDirPaths instead of deprecated baseTaskDirPath
* Use getBaseTaskDirPaths in generate command
The FiniteFirehoseFactory and InputRowParser classes were deprecated in 0.17.0 (#8823) in favor of InputSource & InputFormat. This PR removes the FiniteFirehoseFactory and all its implementations along with classes solely used by them like Fetcher (Used by PrefetchableTextFilesFirehoseFactory). Refactors classes including tests using FiniteFirehoseFactory to use InputSource instead.
Removing InputRowParser may not be as trivial as many classes that aren't deprecated depends on it (with no alternatives), like EventReceiverFirehoseFactory. Hence FirehoseFactory, EventReceiverFirehoseFactory, and Firehose are marked deprecated.
*When running REPLACE queries, the segments which contain no data are dropped (marked as unused). This PR aims to generate tombstones in place of segments which contain no data to mark their deletion, as is the behavior with the native ingestion.
This will cause InsertCannotReplaceExistingSegmentFault to be removed since it was generated if the interval to be marked unused didn't fully overlap one of the existing segments to replace.
If the intermediate handoff period is less than the task duration and there is no new data in the input topic, task will continuously checkpoint the same offsets again and again. This PR fixes that bug by resetting the checkpoint time even when the task receives the same end offset request again.
* merge druid-core, extendedset, and druid-hll into druid-processing to simplify everything
* fix poms and license stuff
* mockito is evil
* allow reset of JvmUtils RuntimeInfo if tests used static injection to override
* Allow users to add additional metadata to ingestion metrics
When submitting an ingestion spec, users may pass a map of metadata
in the ingestion spec config that will be added to ingestion metrics.
This will make it possible for operators to tag metrics with other
metadata that doesn't necessarily line up with the existing tags
like taskId.
Druid clusters that ingest these metrics can take advantage of the
nested data columns feature to process this additional metadata.
* rename to tags
* docs
* tests
* fix test
* make code cov happy
* checkstyle
* discover nested columns when using nested column indexer for schemaless
* move useNestedColumnIndexerForSchemaDiscovery from AppendableIndexSpec to DimensionsSpec
* Kinesis: More robust default fetch settings.
1) Default recordsPerFetch and recordBufferSize based on available memory
rather than using hardcoded numbers. For this, we need an estimate
of record size. Use 10 KB for regular records and 1 MB for aggregated
records. With 1 GB heaps, 2 processors per task, and nonaggregated
records, recordBufferSize comes out to the same as the old
default (10000), and recordsPerFetch comes out slightly lower (1250
instead of 4000).
2) Default maxRecordsPerPoll based on whether records are aggregated
or not (100 if not aggregated, 1 if aggregated). Prior default was 100.
3) Default fetchThreads based on processors divided by task count on
Indexers, rather than overall processor count.
4) Additionally clean up the serialized JSON a bit by adding various
JsonInclude annotations.
* Updates for tests.
* Additional important verify.
* single typed "root" only nested columns now mimic "regular" columns of those types
* incremental index can now use nested column indexer instead of string indexer for discovered columns
* Support Framing for Window Aggregations
This adds support for framing over ROWS
for window aggregations.
Still not implemented as yet:
1. RANGE frames
2. Multiple different frames in the same query
3. Frames on last/first functions
This commit adds a new class `InputStats` to track the total bytes processed by a task.
The field `processedBytes` is published in task reports along with other row stats.
Major changes:
- Add class `InputStats` to track processed bytes
- Add method `InputSourceReader.read(InputStats)` to read input rows while counting bytes.
> Since we need to count the bytes, we could not just have a wrapper around `InputSourceReader` or `InputEntityReader` (the way `CountableInputSourceReader` does) because the `InputSourceReader` only deals with `InputRow`s and the byte information is already lost.
- Classic batch: Use the new `InputSourceReader.read(inputStats)` in `AbstractBatchIndexTask`
- Streaming: Increment `processedBytes` in `StreamChunkParser`. This does not use the new `InputSourceReader.read(inputStats)` method.
- Extend `InputStats` with `RowIngestionMeters` so that bytes can be exposed in task reports
Other changes:
- Update tests to verify the value of `processedBytes`
- Rename `MutableRowIngestionMeters` to `SimpleRowIngestionMeters` and remove duplicate class
- Replace `CacheTestSegmentCacheManager` with `NoopSegmentCacheManager`
- Refactor `KafkaIndexTaskTest` and `KinesisIndexTaskTest`
Refactor DataSource to have a getAnalysis method()
This removes various parts of the code where while loops and instanceof
checks were being used to walk through the structure of DataSource objects
in order to build a DataSourceAnalysis. Instead we just ask the DataSource
for its analysis and allow the stack to rebuild whatever structure existed.
* Zero-copy local deep storage.
This is useful for local deep storage, since it reduces disk usage and
makes Historicals able to load segments instantaneously.
Two changes:
1) Introduce "druid.storage.zip" parameter for local storage, which defaults
to false. This changes default behavior from writing an index.zip to writing
a regular directory. This is safe to do even during a rolling update, because
the older code actually already handled unzipped directories being present
on local deep storage.
2) In LocalDataSegmentPuller and LocalDataSegmentPusher, use hard links
instead of copies when possible. (Generally this is possible when the
source and destination directory are on the same filesystem.)
Changes:
- Limit max batch size in `SegmentAllocationQueue` to 500
- Rename `batchAllocationMaxWaitTime` to `batchAllocationWaitTime` since the actual
wait time may exceed this configured value.
- Replace usage of `SegmentInsertAction` in `TaskToolbox` with `SegmentTransactionalInsertAction`
In a cluster with a large number of streaming tasks (~1000), SegmentAllocateActions
on the overlord can often take very long intervals of time to finish thus causing spikes
in the `task/action/run/time`. This may result in lag building up while a task waits for a
segment to get allocated.
The root causes are:
- large number of metadata calls made to the segments and pending segments tables
- `giant` lock held in `TaskLockbox.tryLock()` to acquire task locks and allocate segments
Since the contention typically arises when several tasks of the same datasource try
to allocate segments for the same interval/granularity, the allocation run times can be
improved by batching the requests together.
Changes
- Add flags
- `druid.indexer.tasklock.batchSegmentAllocation` (default `false`)
- `druid.indexer.tasklock.batchAllocationMaxWaitTime` (in millis) (default `1000`)
- Add methods `canPerformAsync` and `performAsync` to `TaskAction`
- Submit each allocate action to a `SegmentAllocationQueue`, and add to correct batch
- Process batch after `batchAllocationMaxWaitTime`
- Acquire `giant` lock just once per batch in `TaskLockbox`
- Reduce metadata calls by batching statements together and updating query filters
- Except for batching, retain the whole behaviour (order of steps, retries, etc.)
- Respond to leadership changes and fail items in queue when not leader
- Emit batch and request level metrics
* fixes BlockLayoutColumnarLongs close method to nullify internal buffer.
* fixes other BlockLayoutColumnar supplier close methods to nullify internal buffers.
* fix spotbugs
Main changes:
1) Convert SeekableStreamIndexTaskClient to an interface, move old code
to SeekableStreamIndexTaskClientSyncImpl, and add new implementation
SeekableStreamIndexTaskClientAsyncImpl that uses ServiceClient.
2) Add "chatAsync" parameter to seekable stream supervisors that causes
the supervisor to use an async task client.
3) In SeekableStreamSupervisor.discoverTasks, adjust logic to avoid making
blocking RPC calls in workerExec threads.
4) In SeekableStreamSupervisor generally, switch from Futures.successfulAsList
to FutureUtils.coalesce, so we can better capture the errors that occurred
with contacting individual tasks.
Other, related changes:
1) Add ServiceRetryPolicy.retryNotAvailable, which controls whether
ServiceClient retries unavailable services. Useful since we do not
want to retry calls unavailable tasks within the service client. (The
supervisor does its own higher-level retries.)
2) Add FutureUtils.transformAsync, a more lambda friendly version of
Futures.transform(f, AsyncFunction).
3) Add FutureUtils.coalesce. Similar to Futures.successfulAsList, but
returns Either instead of using null on error.
4) Add JacksonUtils.readValue overloads for JavaType and TypeReference.
Currently, a shared lock is acquired only when all other locks are also shared locks.
This commit updates the behaviour and acquires a shared lock only if all locks
of equal or higher priority are either shared locks or are already revoked.
The lock type of locks with lower priority does not matter as they can be revoked.
Eliminates two common sources of noise with Kafka supervisors that have
large numbers of tasks and partitions:
1) Log the report at DEBUG rather than INFO level at each run cycle.
It can get quite large, and can be retrieved via API when needed.
2) Use log4j2.xml to quiet down the org.apache.kafka.clients.consumer.internals
package. Avoids a log message per-partition per-minute as part of seeking
to the latest offset in the reporting thread. In the tasks, where this
sort of logging might be more useful, we have another log message with
the same information: "Seeking partition[%s] to[%s]".
* SeekableStreamSupervisor: Don't enqueue duplicate notices.
Similar goal to #12018, but more aggressive. Don't enqueue a notice at
all if it is equal to one currently in the queue.
* Adjustments from review.
* Update indexing-service/src/test/java/org/apache/druid/indexing/overlord/supervisor/NoticesQueueTest.java
Co-authored-by: Kashif Faraz <kashif.faraz@gmail.com>
Co-authored-by: Kashif Faraz <kashif.faraz@gmail.com>
Changes:
- Add a metric for partition-wise kafka/kinesis lag for streaming ingestion.
- Emit lag metrics for streaming ingestion when supervisor is not suspended and state is in {RUNNING, IDLE, UNHEALTHY_TASKS, UNHEALTHY_SUPERVISOR}
- Document metrics
* Compaction: Fetch segments one at a time on main task; skip when possible.
Compact tasks include the ability to fetch existing segments and determine
reasonable defaults for granularitySpec, dimensionsSpec, and metricsSpec.
This is a useful feature that makes compact tasks work well even when the
user running the compaction does not have a clear idea of what they want
the compacted segments to be like.
However, this comes at a cost: it takes time, and disk space, to do all
of these fetches. This patch improves the situation in two ways:
1) When segments do need to be fetched, download them one at a time and
delete them when we're done. This still takes time, but minimizes the
required disk space.
2) Don't fetch segments on the main compact task when they aren't needed.
If the user provides a full granularitySpec, dimensionsSpec, and
metricsSpec, we can skip it.
* Adjustments.
* Changes from code review.
* Fix logic for determining rollup.
* Support for middle manager less druid, tasks launch as k8s jobs
* Fixing forking task runner test
* Test cleanup, dependency cleanup, intellij inspections cleanup
* Changes per PR review
Add configuration option to disable http/https proxy for the k8s client
Update the docs to provide more detail about sidecar support
* Removing un-needed log lines
* Small changes per PR review
* Upon task completion we callback to the overlord to update the status / locaiton, for slower k8s clusters, this reduces locking time significantly
* Merge conflict fix
* Fixing tests and docs
* update tiny-cluster.yaml
changed `enableTaskLevelLogPush` to `encapsulatedTask`
* Apply suggestions from code review
Co-authored-by: Abhishek Agarwal <1477457+abhishekagarwal87@users.noreply.github.com>
* Minor changes per PR request
* Cleanup, adding test to AbstractTask
* Add comment in peon.sh
* Bumping code coverage
* More tests to make code coverage happy
* Doh a duplicate dependnecy
* Integration test setup is weird for k8s, will do this in a different PR
* Reverting back all integration test changes, will do in anotbher PR
* use StringUtils.base64 instead of Base64
* Jdk is nasty, if i compress in jdk 11 in jdk 17 the decompressed result is different
Co-authored-by: Rahul Gidwani <r_gidwani@apple.com>
Co-authored-by: Abhishek Agarwal <1477457+abhishekagarwal87@users.noreply.github.com>
In clusters with a large number of segments, the duty `MarkAsUnusedOvershadowedSegments`
can take a long very long time to finish. This is because of the costly invocation of
`timeline.isOvershadowed` which is done for every used segment in every coordinator run.
Changes
- Use `DataSourceSnapshot.getOvershadowedSegments` to get all overshadowed segments
- Iterate over this set instead of all used segments to identify segments that can be marked as unused
- Mark segments as unused in the DB in batches rather than one at a time
- Refactor: Add class `SegmentTimeline` for ease of use and readability while using a
`VersionedIntervalTimeline` of segments.
* Remove basePersistDirectory from tuning configs.
Since the removal of CliRealtime, it serves no purpose, since it is
always overridden in production using withBasePersistDirectory given
some subdirectory of the task work directory.
Removing this from the tuning config has a benefit beyond removing
no-longer-needed logic: it also avoids the side effect of empty
"druid-realtime-persist" directories getting created in the systemwide
temp directory.
* Test adjustments to appropriately set basePersistDirectory.
* Remove unused import.
* Fix RATC constructor.
Overlord leader election can sometimes fail due to task lock re-acquisition issues.
This commit solves the issue by failing such tasks and clearing all their locks.
This commit fixes issues with delayed supervisor termination during certain transient states.
Tasks can be created during supervisor termination and left behind since the cleanup may
not consider these newly added tasks.
#12178 added a lock for the entire process of task creation to prevent such dangling tasks.
But it also introduced a deadlock scenario as follows:
- An invocation of `runInternal` is in progress.
- A `stop` request comes, acquires `stateChangeLock` and submit a `ShutdownNotice`
- `runInternal` keeps waiting to acquire the `stateChangeLock`
- `ShutdownNotice` remains stuck in the notice queue because `runInternal` is still running
- After some timeout, the supervisor goes through a forced termination
Fix:
* `SeekableStreamSupervisor.runInternal` - do not try to acquire lock if supervisor is already stopping
* `SupervisorStateManager.maybeSetState` - do not allow transitions from STOPPING state
* Fixing RACE in HTTP remote task Runner
* Changes in the interface
* Updating documentation
* Adding test cases to SwitchingTaskLogStreamer
* Adding more tests
* Fix serialization in TaskReportFileWriters.
For some reason, serializing a Map<String, TaskReport> would omit the
"type" field. Explicitly sending each value through the ObjectMapper
fixes this, because the type information does not get lost.
* Fixes for static analysis.
* Fix race in TaskQueue.notifyStatus.
It was possible for manageInternal to relaunch a task while it was
being cleaned up, due to a race that happens when notifyStatus is
called to clean up a successful task:
1) In a critical section, notifyStatus removes the task from "tasks".
2) Outside a critical section, notifyStatus calls taskRunner.shutdown
to let the task runner know it can clear out its data structures.
3) In a critical section, syncFromStorage adds the task back to "tasks",
because it is still present in metadata storage.
4) In a critical section, manageInternalCritical notices that the task
is in "tasks" and is not running in the taskRunner, so it launches
it again.
5) In a (different) critical section, notifyStatus updates the metadata
store to set the task status to SUCCESS.
6) The task continues running even though it should not be.
The possibility for this race was introduced in #12099, which shrunk
the critical section in notifyStatus. Prior to that patch, a single
critical section encompassed (1), (2), and (5), so the ordering above
was not possible.
This patch does the following:
1) Fixes the race by adding a recentlyCompletedTasks set that prevents
the main management loop from doing anything with tasks that are
currently being cleaned up.
2) Switches the order of the critical sections in notifyStatus, so
metadata store updates happen first. This is useful in case of
server failures: it ensures that if the Overlord fails in the midst
of notifyStatus, then completed-task statuses are still available in
ZK or on MMs for the next Overlord. (Those are cleaned up by
taskRunner.shutdown, which formerly ran first.) This isn't related
to the race described above, but is fixed opportunistically as part
of the same patch.
3) Changes the "tasks" list to a map. Many operations require retrieval
or removal of individual tasks; those are now O(1) instead of O(N)
in the number of running tasks.
4) Changes various log messages to use task ID instead of full task
payload, to make the logs more readable.
* Fix format string.
* Update comment.
Kinesis ingestion requires all shards to have at least 1 record at the required position in druid.
Even if this is satisified initially, resharding the stream can lead to empty intermediate shards. A significant delay in writing to newly created shards was also problematic.
Kinesis shard sequence numbers are big integers. Introduce two more custom sequence tokens UNREAD_TRIM_HORIZON and UNREAD_LATEST to indicate that a shard has not been read from and that it needs to be read from the start or the end respectively.
These values can be used to avoid the need to read at least one record to obtain a sequence number for ingesting a newly discovered shard.
If a record cannot be obtained immediately, use a marker to obtain the relevant shardIterator and use this shardIterator to obtain a valid sequence number. As long as a valid sequence number is not obtained, continue storing the token as the offset.
These tokens (UNREAD_TRIM_HORIZON and UNREAD_LATEST) are logically ordered to be earlier than any valid sequence number.
However, the ordering requires a few subtle changes to the existing mechanism for record sequence validation:
The sequence availability check ensures that the current offset is before the earliest available sequence in the shard. However, current token being an UNREAD token indicates that any sequence number in the shard is valid (despite the ordering)
Kinesis sequence numbers are inclusive i.e if current sequence == end sequence, there are more records left to read.
However, the equality check is exclusive when dealing with UNREAD tokens.
* Refactor Guice initialization
Builders for various module collections
Revise the extensions loader
Injector builders for server startup
Move Hadoop init to indexer
Clean up server node role filtering
Calcite test injector builder
* Revisions from review comments
* Build fixes
* Revisions from review comments
Few indexing tasks register RealtimeMetricsMonitor or TaskRealtimeMetricsMonitor with the process’s MonitorScheduler when they start. These monitors never unregister themselves (they always return true, they'd need to return false to unregister). Each of these monitors emits a set of metrics once every druid.monitoring.emissionPeriod.
As a result, after executing several tasks for a while, Indexer emits metrics of these tasks even after they're long gone.
Proposed Solution
Since one should be able to obtain the last round of ingestion metrics after the task unregisters the monitor, introducing lastRoundMetricsToBePushed variable to keep track of the same and overriding the AbstractMonitor.monitor method in RealtimeMetricsMonitor, TaskRealtimeMetricsMonitor to implement the new logic.
* Add EIGHT_HOUR into possible list of Granularities.
* Add the missing definition.
* fix test.
* Fix another test.
* Stylecheck finally passed.
Co-authored-by: Didip Kerabat <didip@apple.com>
* Mid-level service client and updated high-level clients.
Our servers talk to each other over HTTP. We have a low-level HTTP
client (HttpClient) that is super-asynchronous and super-customizable
through its handlers. It's also proven to be quite robust: we use it
for Broker -> Historical communication over the wide variety of query
types and workloads we support.
But the low-level client has no facilities for service location or
retries, which means we have a variety of high-level clients that
implement these in their own ways. Some high-level clients do a better
job than others. This patch adds a mid-level ServiceClient that makes
it easier for high-level clients to be built correctly and harmoniously,
and migrates some of the high-level logic to use ServiceClients.
Main changes:
1) Add ServiceClient org.apache.druid.rpc package. That package also
contains supporting stuff like ServiceLocator and RetryPolicy
interfaces, and a DiscoveryServiceLocator based on
DruidNodeDiscoveryProvider.
2) Add high-level OverlordClient in org.apache.druid.rpc.indexing.
3) Indexing task client creator in TaskServiceClients. It uses
SpecificTaskServiceLocator to find the tasks. This improves on
ClientInfoTaskProvider by caching task locations for up to 30 seconds
across calls, reducing load on the Overlord.
4) Rework ParallelIndexSupervisorTaskClient to use a ServiceClient
instead of extending IndexTaskClient.
5) Rework RemoteTaskActionClient to use a ServiceClient instead of
DruidLeaderClient.
6) Rework LocalIntermediaryDataManager, TaskMonitor, and
ParallelIndexSupervisorTask. As a result, MiddleManager, Peon, and
Overlord no longer need IndexingServiceClient (which internally used
DruidLeaderClient).
There are some concrete benefits over the prior logic, namely:
- DruidLeaderClient does retries in its "go" method, but only retries
exactly 5 times, does not sleep between retries, and does not retry
retryable HTTP codes like 502, 503, 504. (It only retries IOExceptions.)
ServiceClient handles retries in a more reasonable way.
- DruidLeaderClient's methods are all synchronous, whereas ServiceClient
methods are asynchronous. This is used in one place so far: the
SpecificTaskServiceLocator, so we don't need to block a thread trying
to locate a task. It can be used in other places in the future.
- HttpIndexingServiceClient does not properly handle all server errors.
In some cases, it tries to parse a server error as a successful
response (for example: in getTaskStatus).
- IndexTaskClient currently makes an Overlord call on every task-to-task
HTTP request, as a way to find where the target task is. ServiceClient,
through SpecificTaskServiceLocator, caches these target locations
for a period of time.
* Style adjustments.
* For the coverage.
* Adjustments.
* Better behaviors.
* Fixes.
* Poison StupidPool and fix resource leaks
There are various resource leaks from test setup as well as some
corners in query processing. We poison the StupidPool to start failing
tests when the leaks come and fix any issues uncovered from that so
that we can start from a clean baseline.
Unfortunately, because of how poisoning works,
we can only fail future checkouts from the same pool,
which means that there is a natural race between a
leak happening -> GC occurs -> leak detected -> pool poisoned.
This race means that, depending on interleaving of tests,
if the very last time that an object is checked out
from the pool leaks, then it won't get caught.
At some point in the future, something will catch it,
however and from that point on it will be deterministic.
* Remove various things left over from iterations
* Clean up FilterAnalysis and add javadoc on StupidPool
* Revert changes to .idea/misc.xml that accidentally got pushed
* Style and test branches
* Stylistic woes
The expiry timeout is compared against the current time but the condition is reversed.
This means that as soon as a supervisor task finishes, its partitions are cleaned up,
irrespective of the specified `intermediaryPartitionTimeout` period.
After these changes, the `intermediaryPartitionTimeout` will start getting honored.
Changes
* Fix the condition
* Add tests to verify the new correct behaviour
* Reduce the default expiry timeout from P1D to PT5M
to retain current behaviour in case of default configs.
* Fix flaky KafkaIndexTaskTest.
The testRunTransactionModeRollback case had many race conditions. Most notably,
it would commit a transaction and then immediately check to see that the results
were *not* indexed. This is racey because it relied on the indexing thread being
slower than the test thread.
Now, the case waits for the transaction to be processed by the indexing thread
before checking the results.
* Changes from review.
* Remove null and empty fields from native queries
* Test fixes
* Attempted IT fix.
* Revisions from review comments
* Build fixes resulting from changes suggested by reviews
* IT fix for changed segment size
The web-console (indirectly) calls the Overlord’s GET tasks API to fetch the tasks' summary which in turn queries the metadata tasks table. This query tries to fetch several columns, including payload, of all the rows at once. This introduces a significant memory overhead and can cause unresponsiveness or overlord failure when the ingestion tab is opened multiple times (due to several parallel calls to this API)
Another thing to note is that the task table (the payload column in particular) can be very large. Extracting large payloads from such tables can be very slow, leading to slow UI. While we are fixing the memory pressure in the overlord, we can also fix the slowness in UI caused by fetching large payloads from the table. Fetching large payloads also puts pressure on the metadata store as reported in the community (Metadata store query performance degrades as the tasks in druid_tasks table grows · Issue #12318 · apache/druid )
The task summaries returned as a response for the API are several times smaller and can fit comfortably in memory. So, there is an opportunity here to fix the memory usage, slow ingestion, and under-pressure metadata store by removing the need to handle large payloads in every layer we can. Of course, the solution becomes complex as we try to fix more layers. With that in mind, this page captures two approaches. They vary in complexity and also in the degree to which they fix the aforementioned problems.
Gian Merlino
Clint Wylie
Maytas Monsereenusorn
AmatyaAvadhanula
Kashif Faraz
Karan Kumar
imply-cheddar
George Shiqi Wu
Harini Rajendran
Andreas Maechler
zachjsh
Rishabh Singh
Alexander Saydakov
Abhishek Radhakrishnan
Suneet Saldanha
Tejaswini Bandlamudi
Nicholas Lippis
TSFenwick
Adarsh Sanjeev
Rohan Garg
kaijianding
Laksh Singla
Abhishek Agarwal
Paul Rogers
Churro
somu-imply
Jonathan Wei
Herb Brewer
刘小辉
Didip Kerabat