More specifically, fix the case where a compressed V0 or V1 message is
larger than the producer batch size.
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Apurva Mehta <apurva@confluent.io>, Ismael Juma <ismael@juma.me.uk>
Closes#3356 from hachikuji/KAFKA-5456
Console Consumer help doesn't say that ``--partition`` option needs ``--offset`` otherwise will consume from the end of the partition. This minor fix makes that happen.
Author: Dustin Cote <dustin@confluent.io>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3153 from cotedm/console_consumer_help_fix
Author: Apurva Mehta <apurva@confluent.io>
Reviewers: Jason Gustafson <jason@confluent.io>, Ismael Juma <ismael@juma.me.uk>
Closes#3355 from apurvam/KAFKA-5457-memoryrecordsbuilder-has-room-for-should-account-for-headers
Removed the first two in favour of the latter.
Author: Kamal C <kamal.chandraprakash@gmail.com>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3350 from Kamal15/utilcleanup
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Apurva Mehta <apurva@confluent.io>, Ismael Juma <ismael@juma.me.uk>
Closes#3348 from hachikuji/fix-has-unflushed-synchronization
In the event of a crash, we always restore the data from the backing changelog. So it seems that we don't need to
persist all data to disk by calling flush when committing. Frequent flushing leads to a large number of small files for compaction increasing
compaction pressure. This PR will perform benchmarks to see if there is any performance gain in not calling `flush()` each time we commit.
Author: Bill Bejeck <bill@confluent.io>
Reviewers: Damian Guy <damian.guy@gmail.com>, Eno Thereska <eno.thereska@gmail.com>, Guozhang Wang <wangguoz@gmail.com>
Closes#3211 from bbejeck/MINOR_no_flush_on_commit
Author: Ismael Juma <ismael@juma.me.uk>
Reviewers: Colin P. Mccabe <cmccabe@confluent.io>, Jason Gustafson <jason@confluent.io>
Closes#3339 from ijuma/kafka-5275-admin-client-api-consistency
Increased the timeout from 30sec to 60sec. When running the system tests with packaged Kafka, Connect workers can take about 30seconds to start.
Author: Randall Hauch <rhauch@gmail.com>
Reviewers: Ewen Cheslack-Postava <ewen@confluent.io>
Closes#3344 from rhauch/KAFKA-5450
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Apurva Mehta <apurva@confluent.io>, Ismael Juma <ismael@juma.me.uk>
Closes#3340 from hachikuji/add-random-aborts-to-system-test
Also remove broker-id tags as we generally use them to provide
additional context.
Finally, do a few clean-ups (could not resist).
Author: Ismael Juma <ismael@juma.me.uk>
Reviewers: Guozhang Wang <wangguoz@gmail.com>
Closes#3341 from ijuma/tweak-transaction-metrics
- multi-subtopology tests
- fencing test
- producer fenced bug fix: Streams did not recover correctly from ProducerFencedException
Author: Matthias J. Sax <matthias@confluent.io>
Reviewers: Ismael Juma <ismael@juma.me.uk>, Bill Bejeck <bill@confluent.io>, Guozhang Wang <wangguoz@gmail.com>
Closes#3276 from mjsax/kafka-5361-add-eos-integration-tests-for-streams-api
Publish Javadoc for common.annotation package, which contains
InterfaceStability.
Finally, mark AdminClient classes with `Evolving` instead of `Unstable`.
Author: Ismael Juma <ismael@juma.me.uk>
Reviewers: Colin Mccabe, Gwen Shapira
Closes#3316 from ijuma/kafka-5274-admin-client-javadoc
Headers are only preserved if the new consumer is
used since the old consumer does not support
them.
Add test case to verify the fix and to avoid regression.
Author: Michael Andre Pearce <Michael.Andre.Pearce@me.com>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3322 from michaelandrepearce/KAFKA-5354
And add warning about usage.
Author: Ismael Juma <ismael@juma.me.uk>
Reviewers: Jason Gustafson <jason@confluent.io>
Closes#3330 from ijuma/extended-serializer-javadoc
NetworkClient#disconnect should not erase the connection information. This will allow exponential
backoff to occur.
Author: Colin P. Mccabe <cmccabe@confluent.io>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3309 from cmccabe/disc
Related to https://github.com/apache/kafka/pull/3321
Author: Konstantine Karantasis <konstantine@confluent.io>
Reviewers: Ewen Cheslack-Postava <ewen@confluent.io>
Closes#3326 from kkonstantine/MINOR-Add-tests-for-PluginDesc
Before this patch, the `partitionErrors` was an immutable map. As a result if a single producer had a marker for multiple partitions, and if there were multiple response callbacks for a single append, we would get an `UnsupportedOperationException` in the `writeTxnMarker` handler.
Author: Apurva Mehta <apurva@confluent.io>
Reviewers: Ismael Juma <ismael@juma.me.uk>, Jason Gustafson <jason@confluent.io>
Closes#3313 from apurvam/KAFKA-5438-fix-unsupportedoperationexception-in-writetxnmarker
The default client-id bandwidth quota config properties have been marked deprecated in the doc, but a warning may be useful before the property is removed in a future release.
Author: Rajini Sivaram <rajinisivaram@googlemail.com>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3315 from rajinisivaram/MINOR-deprecate-staticquota
Skip topics that don't have any partitions in zkUtils.getAllPartitions()
Author: Mickael Maison <mickael.maison@gmail.com>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3295 from mimaison/KAFKA-5418
Assign non-null tp as soon as possible once we know the partition. This is
so that if ensureValidRecordSize() throws, the
interceptors.onSendError() call is made with a non-null tp.
Author: Tom Bentley <tbentley@redhat.com>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3280 from tombentley/tp-assign
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Apurva Mehta <apurva@confluent.io>, Ismael Juma <ismael@juma.me.uk>
Closes#3298 from hachikuji/KAFKA-5428
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Apurva Mehta <apurva@confluent.io>, Ismael Juma <ismael@juma.me.uk>
Closes#3300 from hachikuji/KAFKA-5429
When the message copier hangs (like when there is a bug in the client), it ignores the sigterm and doesn't shut down. this leaves the cluster in an unclean state causing future tests to fail.
In this patch we always send SIGKILL when cleaning the node if the process isn't already dead. This is consistent with the other services.
Author: Apurva Mehta <apurva@confluent.io>
Reviewers: Jason Gustafson <jason@confluent.io>
Closes#3308 from apurvam/KAFKA-5437-force-kill-message-copier-on-cleanup
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Ismael Juma <ismael@juma.me.uk>, Apurva Mehta <apurva@confluent.io>, Guozhang Wang <wangguoz@gmail.com>
Closes#3297 from hachikuji/KAFKA-5427
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Guozhang Wang <wangguoz@gmail.com>, Ismael Juma <ismael@juma.me.uk>
Closes#3284 from hachikuji/minor-either-usage-cleanup
In `TransationStateManager`, we reset the pending state if an error occurred while appending to log; this is correct except that for the `TransactionMarkerChannelManager`, as it will retry appending to log and if eventually it succeeded, the transaction metadata's completing transition will throw an IllegalStateException since pending state is None, this will be thrown all the way to the `KafkaApis` and be swallowed.
1. Do not reset the pending state if the append will be retried (as is the case when write the complete transition).
2. A bunch of log4j improvements based the debugging experience. The main principle is to make sure all error codes that is about to sent to the client will be logged, and unnecessary log4j entries to be removed.
3. Also moved some log entries in ReplicationUtils.scala to `trace`: this is rather orthogonal to this PR but I found it rather annoying while debugging the logs.
4. A couple of unrelated bug fixes as pointed by hachikuji and apurvam.
Author: Guozhang Wang <wangguoz@gmail.com>
Reviewers: Apurva Mehta <apurva@confluent.io>, Jason Gustafson <jason@confluent.io>
Closes#3287 from guozhangwang/KHotfix-transaction-coordinator-append-callback
Record `apiThrottleTime` in RequestChannel.
junrao A trivial change. Please review. Thanks.
Author: huxihx <huxi_2b@hotmail.com>
Reviewers: Jun Rao <junrao@gmail.com>
Closes#3265 from huxihx/KAFKA-5405
Author: Jeyhun Karimov <je.karimov@gmail.com>
Reviewers: Matthias J. Sax <matthias@confluent.io>, Guozhang Wang <wangguoz@gmail.com>
Closes#3288 from jeyhunkarimov/KAFKA-4661
This assertion is hard to get right because the system time can roll backward on a host due to NTP (as shown in the ticket), and also because a transaction can start on one host and complete on another. Getting precise clock times across hosts is virtually impossible, and this check makes things fragile.
Author: Apurva Mehta <apurva@confluent.io>
Reviewers: Guozhang Wang <wangguoz@gmail.com>, Jason Gustafson <jason@confluent.io>
Closes#3286 from apurvam/KAFKA-5415-avoid-timestamp-check-in-completeTransition
Originally we assume the task will be created exactly three times (twice upon starting up, once for each thread, and then one more time when rebalancing upon the thread failure). However there is a likelihood that upon starting up more than one rebalance will be triggered, and hence the tasks will be initialized more than 3 times, i.e. there will be more than three hashcodes of the `Transformer` object, causing the `errorInjected` to never be taken and exception never thrown.
The current fix is to use an atomic boolean instead and let threads compete on compare-and-set to make sure exactly one thread will throw exception, and will only throw once.
Without this patch I can reproduce the issue on my local machine with a single core ever 3-5 times; with this patch I have been running successfully for 10+ runs.
Ping mjsax ijuma for reviews.
Author: Guozhang Wang <wangguoz@gmail.com>
Reviewers: Matthias J. Sax <matthias@confluent.io>, Jason Gustafson <jason@confluent.io>
Closes#3275 from guozhangwang/KHotfix-eos-integration-test
This reverts commit d7d1196a0b.
Author: Jason Gustafson <jason@confluent.io>
Reviewers: Ismael Juma <ismael@juma.me.uk>
Closes#3277 from hachikuji/KAFKA-5414
Michal Borowiecki
Jason Gustafson
Dustin Cote
Apurva Mehta
Tom Bentley
Kamal C
Bill Bejeck
Matthias J. Sax
Ismael Juma
Randall Hauch
Ewen Cheslack-Postava
Michael Andre Pearce
Colin P. Mccabe
Konstantine Karantasis
Rajini Sivaram
Mickael Maison
Manikumar Reddy O
Guozhang Wang
huxihx
Jeyhun Karimov