Memory leak in new client API (Java) in 0.8.2.1?

Hi, I'm seeing slow off-heap memory leak in production. I've managed to recreated the scenario in a testing environment - I have one ZooKeeper node and one Kafka Broker (running on same machine). I have one java process which runs a thread which constantly writes to Kafka using 16 KafkaProducer. I

Re: Memory leak in new client API (Java) in 0.8.2.1?

No. Since I run out of memory on the machine, but heap remains low (700mb - 2g) On Sun, Feb 28, 2016 at 7:44 PM Christian Posta wrote: > Do you eventually run OOM? > > On Sunday, February 28, 2016, Asaf Mesika wrote: > > > Hi, > > > > I'm seeing slow

Re: Large Size Error even when the message is small

Can you show your code for sending? On Tue, 1 Mar 2016 at 21:59 Fang Wong wrote: > [2016-02-26 20:33:42,997] INFO Closing socket connection to /x due to > invalid request: Request of length 1937006964 is not valid, [2016-02-26 > 20:33:42,997] INFO Closing socket connection to /10.224.146.58 due

Re: dumping JMX data

If I'm not mistaken jmxtrans does not let you take All metrics beans or just a group of them using wildcard. You are forced to specify the exact bean name which this email shows how cumbersome this is. Also jmxtrans issue an rpc call per bean while using Jolokia you do one post request to get all t

Re: New consumer: OutOfMemoryError: Direct buffer memory

Track the nio direct memory buffer size on your client jvm. Connect to it using VisualVM or Jolokia with halt.io and see where it breaks. This error means your host ran out of memory unless you limit it. On Tue, 12 Apr 2016 at 03:18 Kanak Biscuitwala wrote: > Hi, > > I'm running Kafka's new consu

Re: Spikes in kafka bytes out (while bytes in remain the same)

Where exactly do you get the measurement from? Your broker? Do you have only one? Your producer? Your spark job? On Mon, 11 Apr 2016 at 23:54 Jorge Rodriguez wrote: > We are running a kafka cluster for our real-time pixel processing > pipeline. The data is produced from our pixel servers into ka

Re: Spikes in kafka bytes out (while bytes in remain the same)

n some GC tuning should alleviate this. > > However, even if this is the case, this would not explain though why under > normal operations, the number of bytes out is 2x the number of bytes in. > Since I only have 1 consumer for each topic, I would expect the numbers to > be fairly clos

Re: Does the Kafka Streams DSL support non-Kafka sources/sinks?

I'd stay off the Camel. It's performance is quite low. Up to 5-10 mb/sec it׳s ok but above that it will be your bottleneck. The problem with Camel is that sometime it's Endpoints have special behavior which is hard to understand and debugging it is a mess. We are now migrating away from it. On Fri

Partition size skew using default partitioner (without key)

ith the new Producer API. We've using default partitioner, and we're *not* supplying partition number nor a key, thus it should be round-robin. >From some reason, we see this. I was wondering if someone from the community ever encountered such a behaviour? Thanks! Asaf Mesika Logz.io

Re: kafka + logstash

You should try opening an issue in the github repo of this plugin. Also try writing in the LogStash forum On יום א׳, 19 ביוני 2016 at 20:39 Fahimeh Ashrafy wrote: > Hello all > > I use kafka input and kafka output plugin in logstash. I have high cpu > usage, what can I do to get it better? > log

Re: Partition size skew using default partitioner (without key)

As we continue to track down the cause I'm trying to ping back here in case someone new might have an answer to the question below? On Thu, Jun 16, 2016 at 12:39 PM Asaf Mesika wrote: > Hi, > > We've noticed that we have some partitions receiving more messages than > oth

Re: Partition size skew using default partitioner (without key)

Since the image is now shown, here's a direct link to it: https://s32.postimg.org/xoet3vu2t/image.png On Tue, Jul 5, 2016 at 7:01 AM Asaf Mesika wrote: > As we continue to track down the cause I'm trying to ping back here in > case someone new might have an answer to th

Re: Partition size skew using default partitioner (without key)

flavor - I won't be exposed to this behaviour since you moved to NIO thus have a single socket per broker? On Tue, Jul 5, 2016 at 7:14 AM Asaf Mesika wrote: > Since the image is now shown, here's a direct link to it: > https://s32.postimg.org/xoet3vu2t/image.png > > On Tue,

Re: JMXTrans Monitoring for Individual Topics

Something we use which be helpful: We use jmx2graphite , so we won't have to manually pick and choose metrics - we simply send them all. When all hell breaks loose, you never know in advance which metric you needed to research root cause analysis. On Fri, J

Getting "Illegal batch type" exception on consumers

27;t do it for DefaultRecordBatch - can't figure out why.. To me it seems like a bug. I've posted a JIRA ticket <https://issues.apache.org/jira/browse/KAFKA-7769>, but not comments since Dec 26th, so I though I can ping here as well and get some pointers from the community. Our current work-around is to restart either the server or client, and it solves it. Thanks! Asaf Mesika

Re: Release Kafka 0.9.0.2?

Waiting for the retries fix which doesn't work before 0.9.0.2 - had to apply the Pull Request my self to a forked repo to get our Log Receiver working properly. It's a must in my opinion. On Thu, Dec 1, 2016 at 6:46 PM Ben Osheroff wrote: > +1. At the least an answer regarding timelines would

Re: Detecting when all the retries are expired for a message

There's a critical bug in that section that has only been fixed in 0.9.0.2 which has not been release yet. Without the fix it doesn't really retry. I forked the kafka repo, applied the fix, built it and placed it in our own Nexus Maven repository until 0.9.0.2 will be released. https://github.com/

Re: Processing older records Kafka Consumer

Do you use auto-commit or committing your self? I'm trying to figure out how the offset moved if it was stuck. On Tue, Dec 6, 2016 at 10:28 AM Amit K wrote: > Hi, > > Is there any way to re-consume the older records from Kafka broker with > kafka consumer? > > I am using kafka 0.9.0.0 In one of

Re: Detecting when all the retries are expired for a message

NotEnoughReplicasAfterAppendException > > NotEnoughReplicasException OffsetOutOfRangeException TimeoutException > > UnknownTopicOrPartitionException" > > > > @asaf :My kafka - API version is 0.10.0.1. So I think I should not > > face the issue that you are mention

Re: Processing older records Kafka Consumer

sume them all for last 2 days. > > Will seek() be helpful, like having another application tuned to same topic > and start consuming in it from approx offset that was there 2 days before? > > Thanks for help in advance! > > On Tue, Dec 6, 2016 at 3:35 PM, Asaf Mesika wrote: &g

Re: Topic discovery when supporting multiple kafka clusters

Why not re-use same cluster? You can assign topics to be live only within a specific set of brokers. Thus you have one "bus" for messages, simplifying your applications code and configurations On Mon, Dec 5, 2016 at 9:43 PM Yifan Ying wrote: > Hi, > > Initially, we have only one Kafka cluster sh

Re: Storing Kafka Message JSON to deep storage like S3

We rolled our own since we couldn't (1.5 years ago) find one. The code is quite simple and short. On Tue, Dec 6, 2016 at 1:55 PM Aseem Bansal wrote: > I just meant that is there an existing tool which does that. Basically I > tell it "Listen to all X streams and write them to S3/HDFS at Y path

Re: Kafka and zookeeper stores and mesos env

Off-question a bit - Using the Kafka Mesos framework should save you from handling those questions: https://github.com/mesos/kafka On Thu, Dec 8, 2016 at 2:33 PM Mike Marzo wrote: If i'm running a 5 node zk cluster and a 3 node kafka cluster in dcker on a mesos/marathon environment where my zk

Re: Processing time series data in order

There is a much easier approach: your can route all messages of a given Id to a specific partition. Since each partition has a single writer you get the ordering you wish for. Of course this won't work if your updates occur in different hosts. Also maybe Kafka streams can help shard the based on it

Re: Retries not being run in KafkaProducer

Search the mailing list I wrote about it. I forked and applied the commit that fixed that. I'm waiting for a few months for 0.9.0.2 but still nothing- don't know what is taking so long if such an important bug is in the code. On Tue, 20 Dec 2016 at 22:55 Andrew Clarkson < andrew.clark...@rallyhealt

Re: Apache Kafka integration using Apache Camel

Don't specify Kafka dependencies. Camel will transitively bring it. Otherwise you are causing version conflict. On Mon, 9 Jan 2017 at 14:20 Kamal C wrote: > Can you enable DEBUG logs ? It'll be helpful to debug. > > -- Kamal > > On Mon, Jan 9, 2017 at 5:37 AM, Gupta, Swati wrote: > > > Hello All

Re: Failed to update metadata after 5000 ms

Without the exact log containing the exception it will be hard to help On Wed, 10 May 2017 at 23:34 IT Consultant <0binarybudd...@gmail.com> wrote: > Hi All, > > Currently , i am running TLS enabled multi-node kafka . > > *Version :* 2.11-0.10.1.1 > > *Scenario :* Whenever producer tries to produc