You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@kafka.apache.org by "M. Manna" <ma...@gmail.com> on 2017/07/19 13:11:57 UTC
Broker Startup fail on a 2-node cluster setup
Hello,
After running an emulated 3 node cluster on my local pc, I am now trying to
deploy a 2-node cluster on two of my remote machines. The following is the
server configuration:
broker.id=1
advertised.listeners=PLAINTEXT://0.0.0.0:9092
listener.security.protocol.map=PLAINTEXT:PLAINTEXT,SSL:SSL
num.network.threads=9
num.io.threads=24
socket.send.buffer.bytes=102400
socket.receive.buffer.bytes=102400
socket.request.max.bytes=999999999
log.dirs=/kafka1
num.partitions=1
num.recovery.threads.per.data.dir=1
log.retention.hours=2
log.retention.minutes=15
log.retention.bytes=20971520
log.segment.bytes=10485760
log.roll.hours=1
log.retention.check.interval.ms=300000
offsets.retention.minutes=2880
offsets.retention.check.interval.ms=300000
log.cleanup.policy=delete
zookeeper.connect=available-server-1:2181,available-server-2:2181
zookeeper.connection.timeout.ms=35000
zookeeper.session.timeout.ms=20000
compression.type=gzip
delete.topic.enable=true
kafka.metrics.polling.interval.secs=5
kafka.metrics.reporters=kafka.metrics.KafkaCSVMetricsReporter
kafka.csv.metrics.dir=/tmp/kafka_metrics
kafka.csv.metrics.reporter.enabled=false
controlled.shutdown.max.retries=10
controlled.shutdown.retry.backoff.ms=3000
request.timeout.ms=45000
We are starting the clusters in the following order
1) available-server-1 (execute zookeeper-server-start followed by
kafka-server-start)
2) available-server-2 (execute zookeeper-server-start followed by
kafka-server-start)
The brokers cannot start up and need to be started separately. Is there any
particular order the cluster nodes should be started?
Best Regards,
Re: Broker Startup fail on a 2-node cluster setup
Posted by "M. Manna" <ma...@gmail.com>.
Just realised - it should never have worked or at least put some meaningful
error message.
https://zookeeper.apache.org/doc/trunk/zookeeperStarted.html#sc_InstallingSingleMode
At least 3 to sustain a single point failure.
On 19 July 2017 at 14:21, mostolog@gmail.com <mo...@gmail.com> wrote:
> To be honest, my advice would be to consider docker
>
> - install docker
> - create a swarm with both nodes
> - deploy a zookeeper service with 3 instances
> - deploy a kafka service with 3 instances
>
>
>
> On 19/07/17 15:19, M. Manna wrote:
>
>> Hello,
>>
>> Thanks for the response. So as a workaround, can I do the following:
>>
>> 1) Node 1 - zookeeper 1 + 2 +3 (brokers 1,2,3)
>> 2) Node 2 - zookeeper 4 + 5 + 6 (brokers 4,5,6)
>>
>> Do you recommend this way?
>>
>>
>> Regards,
>>
>>
>> On 19 July 2017 at 14:12, mostolog@gmail.com <mo...@gmail.com> wrote:
>>
>> AFAIK you cant deploy a 2 node cluster, or you'll have split brains issues
>>> "out of the box"
>>>
>>>
>>>
>>> On 19/07/17 15:11, M. Manna wrote:
>>>
>>> Hello,
>>>>
>>>> After running an emulated 3 node cluster on my local pc, I am now trying
>>>> to
>>>> deploy a 2-node cluster on two of my remote machines. The following is
>>>> the
>>>> server configuration:
>>>>
>>>> broker.id=1
>>>> advertised.listeners=PLAINTEXT://0.0.0.0:9092
>>>> listener.security.protocol.map=PLAINTEXT:PLAINTEXT,SSL:SSL
>>>> num.network.threads=9
>>>> num.io.threads=24
>>>> socket.send.buffer.bytes=102400
>>>> socket.receive.buffer.bytes=102400
>>>> socket.request.max.bytes=999999999
>>>> log.dirs=/kafka1
>>>> num.partitions=1
>>>> num.recovery.threads.per.data.dir=1
>>>> log.retention.hours=2
>>>> log.retention.minutes=15
>>>> log.retention.bytes=20971520
>>>> log.segment.bytes=10485760
>>>> log.roll.hours=1
>>>> log.retention.check.interval.ms=300000
>>>> offsets.retention.minutes=2880
>>>> offsets.retention.check.interval.ms=300000
>>>> log.cleanup.policy=delete
>>>> zookeeper.connect=available-server-1:2181,available-server-2:2181
>>>> zookeeper.connection.timeout.ms=35000
>>>> zookeeper.session.timeout.ms=20000
>>>> compression.type=gzip
>>>> delete.topic.enable=true
>>>> kafka.metrics.polling.interval.secs=5
>>>> kafka.metrics.reporters=kafka.metrics.KafkaCSVMetricsReporter
>>>> kafka.csv.metrics.dir=/tmp/kafka_metrics
>>>> kafka.csv.metrics.reporter.enabled=false
>>>> controlled.shutdown.max.retries=10
>>>> controlled.shutdown.retry.backoff.ms=3000
>>>> request.timeout.ms=45000
>>>>
>>>> We are starting the clusters in the following order
>>>>
>>>> 1) available-server-1 (execute zookeeper-server-start followed by
>>>> kafka-server-start)
>>>> 2) available-server-2 (execute zookeeper-server-start followed by
>>>> kafka-server-start)
>>>>
>>>> The brokers cannot start up and need to be started separately. Is there
>>>> any
>>>> particular order the cluster nodes should be started?
>>>>
>>>> Best Regards,
>>>>
>>>>
>>>>
>
Re: Broker Startup fail on a 2-node cluster setup
Posted by "mostolog@gmail.com" <mo...@gmail.com>.
To be honest, my advice would be to consider docker
- install docker
- create a swarm with both nodes
- deploy a zookeeper service with 3 instances
- deploy a kafka service with 3 instances
On 19/07/17 15:19, M. Manna wrote:
> Hello,
>
> Thanks for the response. So as a workaround, can I do the following:
>
> 1) Node 1 - zookeeper 1 + 2 +3 (brokers 1,2,3)
> 2) Node 2 - zookeeper 4 + 5 + 6 (brokers 4,5,6)
>
> Do you recommend this way?
>
>
> Regards,
>
>
> On 19 July 2017 at 14:12, mostolog@gmail.com <mo...@gmail.com> wrote:
>
>> AFAIK you cant deploy a 2 node cluster, or you'll have split brains issues
>> "out of the box"
>>
>>
>>
>> On 19/07/17 15:11, M. Manna wrote:
>>
>>> Hello,
>>>
>>> After running an emulated 3 node cluster on my local pc, I am now trying
>>> to
>>> deploy a 2-node cluster on two of my remote machines. The following is the
>>> server configuration:
>>>
>>> broker.id=1
>>> advertised.listeners=PLAINTEXT://0.0.0.0:9092
>>> listener.security.protocol.map=PLAINTEXT:PLAINTEXT,SSL:SSL
>>> num.network.threads=9
>>> num.io.threads=24
>>> socket.send.buffer.bytes=102400
>>> socket.receive.buffer.bytes=102400
>>> socket.request.max.bytes=999999999
>>> log.dirs=/kafka1
>>> num.partitions=1
>>> num.recovery.threads.per.data.dir=1
>>> log.retention.hours=2
>>> log.retention.minutes=15
>>> log.retention.bytes=20971520
>>> log.segment.bytes=10485760
>>> log.roll.hours=1
>>> log.retention.check.interval.ms=300000
>>> offsets.retention.minutes=2880
>>> offsets.retention.check.interval.ms=300000
>>> log.cleanup.policy=delete
>>> zookeeper.connect=available-server-1:2181,available-server-2:2181
>>> zookeeper.connection.timeout.ms=35000
>>> zookeeper.session.timeout.ms=20000
>>> compression.type=gzip
>>> delete.topic.enable=true
>>> kafka.metrics.polling.interval.secs=5
>>> kafka.metrics.reporters=kafka.metrics.KafkaCSVMetricsReporter
>>> kafka.csv.metrics.dir=/tmp/kafka_metrics
>>> kafka.csv.metrics.reporter.enabled=false
>>> controlled.shutdown.max.retries=10
>>> controlled.shutdown.retry.backoff.ms=3000
>>> request.timeout.ms=45000
>>>
>>> We are starting the clusters in the following order
>>>
>>> 1) available-server-1 (execute zookeeper-server-start followed by
>>> kafka-server-start)
>>> 2) available-server-2 (execute zookeeper-server-start followed by
>>> kafka-server-start)
>>>
>>> The brokers cannot start up and need to be started separately. Is there
>>> any
>>> particular order the cluster nodes should be started?
>>>
>>> Best Regards,
>>>
>>>
Re: Broker Startup fail on a 2-node cluster setup
Posted by "M. Manna" <ma...@gmail.com>.
Hello,
Thanks for the response. So as a workaround, can I do the following:
1) Node 1 - zookeeper 1 + 2 +3 (brokers 1,2,3)
2) Node 2 - zookeeper 4 + 5 + 6 (brokers 4,5,6)
Do you recommend this way?
Regards,
On 19 July 2017 at 14:12, mostolog@gmail.com <mo...@gmail.com> wrote:
> AFAIK you cant deploy a 2 node cluster, or you'll have split brains issues
> "out of the box"
>
>
>
> On 19/07/17 15:11, M. Manna wrote:
>
>> Hello,
>>
>> After running an emulated 3 node cluster on my local pc, I am now trying
>> to
>> deploy a 2-node cluster on two of my remote machines. The following is the
>> server configuration:
>>
>> broker.id=1
>> advertised.listeners=PLAINTEXT://0.0.0.0:9092
>> listener.security.protocol.map=PLAINTEXT:PLAINTEXT,SSL:SSL
>> num.network.threads=9
>> num.io.threads=24
>> socket.send.buffer.bytes=102400
>> socket.receive.buffer.bytes=102400
>> socket.request.max.bytes=999999999
>> log.dirs=/kafka1
>> num.partitions=1
>> num.recovery.threads.per.data.dir=1
>> log.retention.hours=2
>> log.retention.minutes=15
>> log.retention.bytes=20971520
>> log.segment.bytes=10485760
>> log.roll.hours=1
>> log.retention.check.interval.ms=300000
>> offsets.retention.minutes=2880
>> offsets.retention.check.interval.ms=300000
>> log.cleanup.policy=delete
>> zookeeper.connect=available-server-1:2181,available-server-2:2181
>> zookeeper.connection.timeout.ms=35000
>> zookeeper.session.timeout.ms=20000
>> compression.type=gzip
>> delete.topic.enable=true
>> kafka.metrics.polling.interval.secs=5
>> kafka.metrics.reporters=kafka.metrics.KafkaCSVMetricsReporter
>> kafka.csv.metrics.dir=/tmp/kafka_metrics
>> kafka.csv.metrics.reporter.enabled=false
>> controlled.shutdown.max.retries=10
>> controlled.shutdown.retry.backoff.ms=3000
>> request.timeout.ms=45000
>>
>> We are starting the clusters in the following order
>>
>> 1) available-server-1 (execute zookeeper-server-start followed by
>> kafka-server-start)
>> 2) available-server-2 (execute zookeeper-server-start followed by
>> kafka-server-start)
>>
>> The brokers cannot start up and need to be started separately. Is there
>> any
>> particular order the cluster nodes should be started?
>>
>> Best Regards,
>>
>>
>
Re: Broker Startup fail on a 2-node cluster setup
Posted by "mostolog@gmail.com" <mo...@gmail.com>.
AFAIK you cant deploy a 2 node cluster, or you'll have split brains
issues "out of the box"
On 19/07/17 15:11, M. Manna wrote:
> Hello,
>
> After running an emulated 3 node cluster on my local pc, I am now trying to
> deploy a 2-node cluster on two of my remote machines. The following is the
> server configuration:
>
> broker.id=1
> advertised.listeners=PLAINTEXT://0.0.0.0:9092
> listener.security.protocol.map=PLAINTEXT:PLAINTEXT,SSL:SSL
> num.network.threads=9
> num.io.threads=24
> socket.send.buffer.bytes=102400
> socket.receive.buffer.bytes=102400
> socket.request.max.bytes=999999999
> log.dirs=/kafka1
> num.partitions=1
> num.recovery.threads.per.data.dir=1
> log.retention.hours=2
> log.retention.minutes=15
> log.retention.bytes=20971520
> log.segment.bytes=10485760
> log.roll.hours=1
> log.retention.check.interval.ms=300000
> offsets.retention.minutes=2880
> offsets.retention.check.interval.ms=300000
> log.cleanup.policy=delete
> zookeeper.connect=available-server-1:2181,available-server-2:2181
> zookeeper.connection.timeout.ms=35000
> zookeeper.session.timeout.ms=20000
> compression.type=gzip
> delete.topic.enable=true
> kafka.metrics.polling.interval.secs=5
> kafka.metrics.reporters=kafka.metrics.KafkaCSVMetricsReporter
> kafka.csv.metrics.dir=/tmp/kafka_metrics
> kafka.csv.metrics.reporter.enabled=false
> controlled.shutdown.max.retries=10
> controlled.shutdown.retry.backoff.ms=3000
> request.timeout.ms=45000
>
> We are starting the clusters in the following order
>
> 1) available-server-1 (execute zookeeper-server-start followed by
> kafka-server-start)
> 2) available-server-2 (execute zookeeper-server-start followed by
> kafka-server-start)
>
> The brokers cannot start up and need to be started separately. Is there any
> particular order the cluster nodes should be started?
>
> Best Regards,
>