The only thing that I find very weird is the fact that brokers that are
dead are still part of the ISR set for hours... and are basically not
removed. Note this is not constantly the case, most of the dead brokers are
properly removed and it is really just in a few cases. I am not sure why
this
Will definitely take a thread dump! So, far its been running fine.
-Jacob
On Wed, Oct 15, 2014 at 8:40 PM, Jun Rao jun...@gmail.com wrote:
If you see the hanging again, it would be great if you can take a thread
dump so that we know where it is hanging.
Thanks,
Jun
On Tue, Oct 14, 2014
Hi,
I’m trying to make a request for offset information from my broker, and I get a
kafka.common.UnknownException as the result.
I’m trying to use the Simple Consumer API
val topicAndPartition = new TopicAndPartition(“topic3”, 0)
val requestInfo = new
Just note that this is not a universal solution. Many use-cases care
about which partition you end up writing to since partitions are used
to... well, partition logical entities such as customers and users.
On Wed, Oct 15, 2014 at 9:03 PM, Jun Rao jun...@gmail.com wrote:
Kyle,
What you
I assume the messages themselves contain the timestamp?
If you use Flume, you can configure a Kafka source to pull data from
Kafka, use an interceptor to pull the date out of your message and
place it in the event header and then the HDFS sink can write to a
partition based on the timestamp.
Check out Camus. It was built to do parallel loads from Kafka into time
bucketed directories in HDFS.
On Oct 16, 2014, at 9:32 AM, Gwen Shapira gshap...@cloudera.com wrote:
I assume the messages themselves contain the timestamp?
If you use Flume, you can configure a Kafka source to pull
I didn't realize that anyone used partitions to logically divide a topic.
When would that be preferable to simply having a separate topic? Isn't this
a minority case?
On Thu, Oct 16, 2014 at 7:28 AM, Gwen Shapira gshap...@cloudera.com wrote:
Just note that this is not a universal solution.
Yes, you should be able to upgrade seamlessly.
On Wed, Oct 15, 2014 at 10:07 PM, Otis Gospodnetic
otis.gospodne...@gmail.com wrote:
Hi,
Some of our SPM users who are eager to monitor their Kafka 0.8.x clusters
with SPM are asking us whether the upgrade to 0.8.2 from 0.8.1 will be
seamless.
In other words, if I change the number of partitions, can I restart the
brokers one at a time so that I can continue processing data?
Changing the # of partitions is an online operation and doesn't require
restarting the brokers. However, any other configuration (with the
exception of a few
Is there a known issue in the 0.8.0 version that was
fixed later on? What can I do to diagnose/fix the situation?
Yes, quite a few bugs related to this have been fixed since 0.8.0. I'd
suggest upgrading to 0.8.1.1
On Wed, Oct 15, 2014 at 11:09 PM, Jean-Pascal Billaud j...@tellapart.com
wrote:
Hi,
I have a question about 'topic.metadata.refresh.interval.ms' configuration.
As I know, the default value of it is 10 minutes.
Does it means that producer will change the partition at every 10 minutes?
What I am experiencing is producer does not change to another
partition at every 10 minutes.
Hi,
I'm trying to monitor the kafka connection on the consumer side. In other
words, if the broker cluster is unavailable (or zookeer dies), I would like
to know about that problem as soon as possible.
Unfortunately, I didn't find anything useful to achieve that when using
kafka library.
Are
hi,
How do I read N items from a topic? I also would like to do this for a
consumer group, so that each consumer can specify an N number of tuples to
read, and each consumer reads distinct tuples.
Thanks,
Josh
Thanks Joseph. I built the javadoc but its incomplete.
Where can I find the code itself for classes like KafkaStream,
MessageAndOffset, CosumerConnector etc?
On Wed, Oct 15, 2014 at 11:10 AM, Joseph Lawson jlaw...@roomkey.com wrote:
You probably have to build your own right now. Check out
It may be a minority, I can't tell yet. But in some apps we need to know that
a consumer, who is assigned a single partition, will get all data about a
subset of users.
This is way more flexible than multiple topics since we still have the benefits
of partition reassignment, load balancing
Using the high level consumer, each consumer in the group can call iter.next ()
in a loop until they get the number of messages you need.
—
Sent from Mailbox
On Thu, Oct 16, 2014 at 10:18 AM, Josh J joshjd...@gmail.com wrote:
hi,
How do I read N items from a topic? I also would like to do
KafkaStream and MessageAndOffset are Scala classes, so you'll find them
under the scaladocs. The ConsumerConnector interface should show up in
the javadocs with good documentation coverage. Some classes like
MessageAndOffset are so simple (just compositions of other data) that
they aren't going to
Josh,
The consumer's API doesn't allow you to specify N messages, but you can
invoke iter.next() as Gwen suggested and count the messages. Note that the
iterator can block if you have less than N messages so you will have to
careful design around it. The new consumer's API provides a non blocking
Another JIRA that will be nice to include as part of 0.8.2-beta is
https://issues.apache.org/jira/browse/KAFKA-1481 that fixes the mbean
naming. Looking for people's thoughts on 2 things here -
1. How do folks feel about doing a 0.8.2-beta release right now and 0.8.2
final 4-5 weeks later?
2. Do
A topic.metadata.refresh.interval.ms of 10 mins means that the producer
will take 10 mins to detect new partitions. So newly added or reassigned
partitions might not get data for 10 mins. In general, if you're still at
prototyping stages, I'd recommend using the new producer available on kafka
If you want to know if the Kafka and zookeeper cluster is healthy or not,
you'd want to monitor the cluster directly. Here are pointers for
monitoring the Kafka brokers -
http://kafka.apache.org/documentation.html#monitoring
Thanks,
Neha
On Thu, Oct 16, 2014 at 3:09 AM, Alex Objelean
Do you see any errors on the broker?
Are you sure that the consumer's fetch offset is set higher than the
largest message in your topic? It should be higher than message.max.bytes
on the broker (which defaults to 1MB).
On Thu, Oct 16, 2014 at 3:56 AM, Magnus Vojbacke
Knowing that the partitioning is consistent for a given key means that
(apart from other benefits) a given consumer only deals with a partition of
the keyspace. So if you are in a system with tens of millions of users each
consumer only has to store state on a small number of them with
+1 for a 0.8.2-beta next week and 0.8.2.0 final 4-5 weeks later.
I agree to the tickets you brought up to have in 0.8.2-beta and also
https://issues.apache.org/jira/browse/KAFKA-1493 for lz4 compression.
/***
Joe Stein
Founder, Principal Consultant
Big
The OffsetRequest can only be answered by the leader of the partition. Did
you connect the SimpleConsumer to the leader broker? If not, you need to
use TopicMetadataRequest to find out the leader broker first.
Thanks,
Jun
On Thu, Oct 16, 2014 at 3:56 AM, Magnus Vojbacke
Which version of ZK are you using? Also, see
https://cwiki.apache.org/confluence/display/KAFKA/FAQ#FAQ-Whyaretheremanyrebalancesinmyconsumerlog
?
Thanks,
Jun
On Thu, Oct 16, 2014 at 3:29 PM, Hari Gorak hari.go...@rediffmail.com
wrote:
Project: Kafka
Issue Type: Bug
Components: consumer
26 matches
Mail list logo