Re: Is kafka 8 and LeaderNotAvailableError

2015-10-11 Thread Prabhjot Bharaj
Wow. Nice. So, what all did you change in your cluster?? Regards, Prabhjot On Oct 11, 2015 7:58 PM, "David Montgomery" wrote: > Its working great now. Thanks > > On Sat, Oct 10, 2015 at 3:35 AM, Prabhjot Bharaj > wrote: > > > Zoo.cfg looks OK to me > > > > Please post your reply whether it

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-11 Thread David Montgomery
Its working great now. Thanks On Sat, Oct 10, 2015 at 3:35 AM, Prabhjot Bharaj wrote: > Zoo.cfg looks OK to me > > Please post your reply whether it worked after changing those values > > If not, we'll figure out something > > Also, I forgot to tell earlier, please check the size of that pa

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread Prabhjot Bharaj
Zoo.cfg looks OK to me Please post your reply whether it worked after changing those values If not, we'll figure out something Also, I forgot to tell earlier, please check the size of that partition's directory on both the replicas. If the size differs by a substantial amount, it clears one thin

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
Thanks for the help :) tickTime=3000 initLimit=10 syncLimit=5 dataDir=/var/lib/zookeeper clientPort=2181 server.1=45.55.xxx.xxx:2888:3888 server.3=104.236.xxx.xxx:2888:3888 server.2=192.241.xxx.xxx:2888:3888 On Sat, Oct 10, 2015 at 3:02 AM, Prabhjot Bharaj wrote: > Hi, > > The error is due to

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread Prabhjot Bharaj
Hi, The error is due to some kind of timeout happening, during fetching of metadata. Random successes and failures can also be explained, if we come to know which timeout parameter is the culprit Default value for replica.fetch.wait.max.ms is 500. Try changing it to some higher value Although I'v

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
Hi I updated the metadata.broker.list=104.236.xxx.xxx:9092,104.236.xxx.xxx:9092 in producer.properties. When I run the code I get random fails and success success: [ProduceResponse(topic='topic-test-production', partition=0, error=0, offset=1)] fail: kafka.common.LeaderNotAvailableError: TopicMet

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
PS I see this in one of the kafka log files Thanks [2015-10-09 13:41:52,873] WARN [ReplicaFetcherThread-0-21441], Error in fetch Name: FetchRequest; Version: 0; CorrelationId: 58; ClientId: ReplicaFetcherThread-0-21441; ReplicaId: 26680; MaxWait: 500 ms; MinBytes: 1 bytes; RequestInfo: [topic-de

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
Is this a total failure on kafka or python kakfa-python? I am using ubuntu 14.04 on digital ocean and zk 3.4.6 Using UFW all zk and kafka servers have access I added my ip address from home to kafka and zk servers using UFW so there should be no filewall issues Below is my kafka server.propertie

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread Prabhjot Bharaj
Hi, Your metadata response is empty. Here is an output for my kafka cluster: *>>> kafka = KafkaClient(['96.a.a.a:9092','96.b.b.b:9092','96.c.c.c:9092','96.d.d.d:9092','96.e.e.e:9092'])* *2015-10-09 21:16:53,575.575.392961502:kafka.client:140735282348800:DEBUG:8705:Request 1: []* *2015-10-09 21:

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
ps here is the output the I add logging to the code import logging logging.basicConfig( format='%(asctime)s.%(msecs)s:%(name)s:%(thread)d:%(levelname)s:%(process)d:%(message)s', level=logging.DEBUG ) 2015-10-09 20:59:33,656.656.464099884:kafka.client:139860093155072:DEBUG:25285:Request 1: []

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
from kafka import SimpleProducer from kafka import KafkaClient kafka_host_list = ['192.241.xxx.xxx:9092','45.55.xxx.xxx:9092','107.170.xxx.xxx:9092'] kafka = KafkaClient(kafka_host_list) producer = SimpleProducer(kafka, async=False) test_topic = 'topic-test-production' test_payload = 'test' print p

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread Prabhjot Bharaj
Hi, Please share the code snippet for Kafka python Regards, Prabhjot On Oct 9, 2015 4:30 PM, "David Montgomery" wrote: > This is the python error > > kafka.common.LeaderNotAvailableError: > TopicMetadata(topic='topic-test-production', error=5, partitions=[]) > > > On Fri, Oct 9, 2015 at 6:55 PM

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
This is the python error kafka.common.LeaderNotAvailableError: TopicMetadata(topic='topic-test-production', error=5, partitions=[]) On Fri, Oct 9, 2015 at 6:55 PM, David Montgomery wrote: > Well i masked my true domain but the hostnames are valid and do work. I > do see the topics creates in

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
Well i masked my true domain but the hostnames are valid and do work. I do see the topics creates in ZK brokers/topics/topic-test-production/partisions/0/state {"controller_epoch":10,"leader":26665,"version":1,"leader_epoch":2,"isr":[26665]} brokers/topics/topic-test-production/partisions/1/state

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread Prabhjot Bharaj
Hi, Is this a valid zookeeper connection string 1.zk.do.production.test:2181,2.zk.do.production.test:2181,3.zk.do.production.test:2181 Is 2.zk.do.production.test or 3.zk.do.production.test a valid hostname?? Regards, Prabhjot On Oct 9, 2015 1:37 PM, "David Montgomery" wrote: > PS I added one

Re: Is kafka 8 and LeaderNotAvailableError

2015-10-09 Thread David Montgomery
PS I added one more server now 3 kafka servers. still kafka is not working. Is there a known bug? On Fri, Oct 9, 2015 at 3:23 PM, David Montgomery wrote: > i, > > I have 3 ZK servers and 2 kafka servers with 2 partitions. > > I am using kafka_2.11-0.8.2.1 > > Here is how I am creating a topic: