kafka message expiry question
hi, as far as i understand, log retention time in kafka will delete message that's older than the retention time. i'm wonder what it implies for consumer since i'm using simple consumer to manage offsets in predefined consumer group. say i have a list of messages for a partition of topic: 1,2,3,4,5 are the message (offsets) associated with the partition in current time. if message 1,2,3 expired earlier and only 4,5 are left, does that mean consumer can only consume 4,5 and need a way to detect 1,2,3 has expired and make sure it never reads before the earliest offset for the partition ? if all 1,2,3,4,5 are expired, it seems the offset will become 0. i assume in this case consumer need to reset its consumed offsets to 0 for the consumer group. Thanks, Weide
Re: offset commit api
Thanks Joel. Do you know if I'm using consumer group string On Mon, Aug 4, 2014 at 3:16 PM, Joel Koshy jjkosh...@gmail.com wrote: Weide, 0.8.1.1 does not support offsets storage in Kafka. The brokers do support offset commit requests/fetches but simply forward to ZooKeeper - you can issue the offset commit and fetch requests to any broker. Kafka-backed consumer offsets is currently in trunk and will be released in 0.8.2. Thanks, Joel On Mon, Aug 04, 2014 at 02:57:02PM -0700, Weide Zhang wrote: Hi It seems to me that 0.8.1.1 doesn't have the ConsumerMetadata API. So what broker I should choose in order to commit and fetch offset information ? Shall I use zookeeper for offset to manage it manually instead ? Thanks, Weide On Sun, Aug 3, 2014 at 4:34 PM, Weide Zhang weo...@gmail.com wrote: Hi, I'm reading the offset management on the API link. https://cwiki.apache.org/confluence/display/KAFKA/A+Guide+To+The+Kafka+Protocol#AGuideToTheKafkaProtocol-OffsetCommit/FetchAPI I have a couple of questions regarding using the offset fetch and commit API in 0.8.1.1 ? 1. Is the new offset commit and fetch api usable in use in 0.8.1.1 ? Does 0.8.1.1 already support offset coordinator ? 2. what's the difference between old offsetrequest and new offsetfetchrequest ? It seems to me that the new api support per consumer group offset management fetch while old api doesn't. Also, what's the purpose of using a timestamp parameter in the fetch request ? 3. In 0.8.1.1, the OffsetCommitRequest uses OffsetMetadataAndError, could you tell me what's the purpose of the error parameter and metadata parameter in the request ? 4. Can I assume the offset management is purely independent of message consumption ? In other words, if I use a simple consumer to fetch message with random client id, can I still manually set some consumer group along with offset in the offset commit message ? Is that allowed ? Thanks, Weide
Re: offset commit api
Hi It seems to me that 0.8.1.1 doesn't have the ConsumerMetadata API. So what broker I should choose in order to commit and fetch offset information ? Shall I use zookeeper for offset to manage it manually instead ? Thanks, Weide On Sun, Aug 3, 2014 at 4:34 PM, Weide Zhang weo...@gmail.com wrote: Hi, I'm reading the offset management on the API link. https://cwiki.apache.org/confluence/display/KAFKA/A+Guide+To+The+Kafka+Protocol#AGuideToTheKafkaProtocol-OffsetCommit/FetchAPI I have a couple of questions regarding using the offset fetch and commit API in 0.8.1.1 ? 1. Is the new offset commit and fetch api usable in use in 0.8.1.1 ? Does 0.8.1.1 already support offset coordinator ? 2. what's the difference between old offsetrequest and new offsetfetchrequest ? It seems to me that the new api support per consumer group offset management fetch while old api doesn't. Also, what's the purpose of using a timestamp parameter in the fetch request ? 3. In 0.8.1.1, the OffsetCommitRequest uses OffsetMetadataAndError, could you tell me what's the purpose of the error parameter and metadata parameter in the request ? 4. Can I assume the offset management is purely independent of message consumption ? In other words, if I use a simple consumer to fetch message with random client id, can I still manually set some consumer group along with offset in the offset commit message ? Is that allowed ? Thanks, Weide
Re: kafka simpleconsumer question
Hi Guozhang, Just a follow up question. Is one simpleconsumer instance can be shared by multiple threads and fetch data from multiple topics and commit offsets ? It seems all the implementation are synchronized which means it can be shared between multiple threads ? Is my understanding correct ? Thanks, Weide On Tue, Jul 1, 2014 at 5:36 PM, Guozhang Wang wangg...@gmail.com wrote: Hi Weide, 1. The old consumer is still depending on the ZK no matter if you use ZK for offsets or not, since it depends on ZK for group manage anyways. In the new consumer (0.9) we have completely the ZK dependance. 2. Yes, it queries the broker for leader info. 3. You can use a single consumer for multiple topics, but a single consumer can only talk to a single broker; if your topic spans across multiple brokers you would also need multiple simple consumers. Guozhang On Tue, Jul 1, 2014 at 5:14 PM, Weide Zhang weo...@gmail.com wrote: Hi , Just want to ask some basic question about kafka simple consumer. 1. if I'm using simple consumer and doesn't really depend on zookeeper to manage partition offset. (application manage offset themselves). Will that remove the zookeeper dependency for consumer ? 2. if zookeeper dies, will simple consumer still able to get partition leadership information from broker itself ? Or it indirectly uses zookeeper to find out the partition leadership for particular topics ? 3. does simpleconsumer api have topicfilter logic similar to high level consumer provides ? Or I have to create multiple simpleconsumer for multiple topics themselves explicitly ? Thanks a lot, Weide -- -- Guozhang
kafka simpleconsumer question
Hi , Just want to ask some basic question about kafka simple consumer. 1. if I'm using simple consumer and doesn't really depend on zookeeper to manage partition offset. (application manage offset themselves). Will that remove the zookeeper dependency for consumer ? 2. if zookeeper dies, will simple consumer still able to get partition leadership information from broker itself ? Or it indirectly uses zookeeper to find out the partition leadership for particular topics ? 3. does simpleconsumer api have topicfilter logic similar to high level consumer provides ? Or I have to create multiple simpleconsumer for multiple topics themselves explicitly ? Thanks a lot, Weide
topics load balancing within a consumer group
Hi, I have a question regarding load balancing within a consumer group. Say I have a consumer group of 4 consumers which subscribe to 4 topics , each of which have one partition. Will there be rebalancing happening on topic level ? Or I will expect consumer 1 have all the data ? Weide
kafka 0.8.1.1 works under java 1.6 ?
Hi, According to the Kafka documentation, seems Java 1.7 is recommended. But for our production environment we are still using java 1.6. Will that be a problem of using java 1.6 and use Kafka 0.8.1.1 ? Thanks a lot, Weide
Re: What happens to Kafka when ZK lost its quorum?
Hi Guozhang, In worst case zookeeper dies for say 1 hour and come back up, will things still be recovered automatically after 1 hour ? Weide On Wed, May 14, 2014 at 8:28 AM, Guozhang Wang wangg...@gmail.com wrote: In 0.8, the servers and consumers are heavily dependent on ZK to function. With ZK down, the servers cannot manage replicas and consumers cannot assign partitions within the group. In 0.9 we are going to remove the ZK dependency from consumer clients, but Kafka servers would still be dependent on ZK. Guozhang On Tue, May 13, 2014 at 6:52 AM, Connie Yang cybercon...@gmail.com wrote: Hi all, Can Kafka producers, brokers and consumers still be processing messages and functioning in their normal states if Zookeeper lost its quorum? Thanks, Connie -- -- Guozhang
zookeeper down
Can kafka survive when zookeeper is down and not connectable ? Will the consumer or producer still work in that case ? Weide
Re: question about mirror maker
Hi Todd, Thanks for your answer. with regard to fail over for mirror maker, does that mean if i have 4 mirror maker running in different machines with same consumer group, it will auto load balance if one of the mirror maker fails ? Also, it looks to prevent mirror maker commit wrong (consumer work but not producer) due to cross data center network issue, mirror maker need to be placed along with the target cluster so that this scenario is minimized ? On Sat, May 10, 2014 at 11:39 PM, Todd Palino tpal...@linkedin.com wrote: Well, if you have a cluster in each datacenter, all with the same topics, you can¹t just mirror the messages between them, as you will create a loop. The way we do it is to have a ³local² cluster and an ³aggregate² cluster. The local cluster has the data for only that datacenter. Then we run mirror makers that copy the messages from each of the local clusters into the aggregate cluster. Everything produces into the local clusters, and nothing produces into the aggregate clusters. In general, consumers consume from the aggregate cluster (unless they specifically want only local data). The mirror maker is as fault tolerant as any other consumer. That is, if a mirror maker goes down, the others configured with the same consumer group (we generally run at least 4 for any mirror maker, sometimes up to 10) will rebalance and start back up from the last committed offset. What you need to watch out for is if the mirror maker is unable to produce messages, for example, if the network goes down. If it can still consume messages, but cannot produce them, you will lose messages as the consumer will continue to commit offsets with no knowledge that the producer is failing. -Todd On 5/8/14, 11:20 AM, Weide Zhang weo...@gmail.com wrote: Hi, I have a question about mirror maker. say I have 3 data centers each producing topic 'A' with separate kafka cluster running. if 3 of the data need to be kept in sync with each other, shall i create 3 mirror maker in each data center to get the data from the other two ? also, it mentioned that mirror making is not fault tolerant ? so what will be the behavior of mirror consumer if it went down due to network and back up ? do they catch up with last offset from which they last mirror ? If so, is it enabled by default or I have to configure ? Thanks a lot, Weide
question about mirror maker
Hi, I have a question about mirror maker. say I have 3 data centers each producing topic 'A' with separate kafka cluster running. if 3 of the data need to be kept in sync with each other, shall i create 3 mirror maker in each data center to get the data from the other two ? also, it mentioned that mirror making is not fault tolerant ? so what will be the behavior of mirror consumer if it went down due to network and back up ? do they catch up with last offset from which they last mirror ? If so, is it enabled by default or I have to configure ? Thanks a lot, Weide