kafka-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From William Reynolds <william.reyno...@instaclustr.com>
Subject Re: Kafka topic partition distributing evenly on disks
Date Fri, 07 Aug 2020 11:00:55 GMT
Hi Péter,
Sounds like time to reassign the partitions you have across all the
brokers/data dirs using the instructions from here
https://kafka.apache.org/documentation/#basic_ops_automigrate. That
assumes that your partition strategy has somewhat evenly filled your
partitions and given it may move all the partitions it could be a bit
intensive so be sure to use the throttle option.
Cheers
William

On 07/08/2020, Péter Nagykátai <st4r.f1sch@gmail.com> wrote:
> Hello everybody,
>
> Thank you for the detailed answers. My issue is partly answered here:
>
>
>
>
> *This rule also applies to disk-level, which means that when a set
> ofpartitions assigned to a specific broker, each of the disks will get
> thesame number of partitions without considering the load of disks at
> thattime.*
>
>  I admit, I didn't provide enough info either.
>
> So my problem is that an existing topic got a huge surge of events for this
> week. I knew that'll happen and I modified the partition count.
> Unfortunately, it occurred to me a bit later, that I'll likely need some
> extra disk space. So I added an extra disk to each broker. The thing I
> didn't know, that Kafka won't evenly distribute the partitions on the
> disks.
> So the question still remains:
>  Is there any way to have Kafka evenly distribute data on its disks?
> Also, what options do I have *after *I'm in the situation I described
> above? (preferably without deleting the topic)
>
> Thanks!
>
> On Fri, Aug 7, 2020 at 12:00 PM Yingshuan Song <songyingshuan@gmail.com>
> wrote:
>
>> Hi Peter,
>> Agreed with Manoj and Vinicius, i think those rules led to this result :
>>
>> 1)the partitions of a topic - N and replication number - R determine the
>> real partition-replica count of this topic, which is N * R;
>> 2)   kafka can distribute partitions evenly among brokers, but it is
>> based
>> on the broker count when the topic was created, this is important.
>> If we create a topic (N - 4, R - 3) in a kafka cluster which contains 3
>> kafka brokers, then 4 * 3 / 3 = 4 partitions will be assigned to each
>> broker.
>> But if a new broker was added into this cluster and another topic (N - 4,
>> R
>> - 3) need to be created, then 4 * 3 / 4 = 3 partitions will be assigned
>> to
>> each broker.
>> Kafka will not assign all those partitions to the new added broker even
>> though it is idle and i think this is a shortcoming of kafka.
>> This rule also applies to disk-level, which means that when a set of
>> partitions assigned to a specific broker, each of the disks will get the
>> same number of partitions without considering the load of disks at that
>> time.
>> 3) when producer send records to topics, how to chose partiton : 3-1) if
>> a
>> record has a key, then the partition number calculate according to the
>> key;
>> 3-2) if  records have no keys, then those records will be sent to each
>> partition in turns. So, if there are lots of records with the same key,
>> and
>> those records will be sent to the same partition, and may take up a lot
>> of
>> disk space.
>>
>>
>> hope this helps
>>
>> Vinicius Scheidegger <vinicius.scheidegger@gmail.com> 于2020年8月7日周五
>> 上午6:10写道:
>>
>> > Hi Peter,
>> >
>> > AFAIK, everything depends on:
>> >
>> > 1) How you have configured your topic
>> >   a) number of partitions (here I understand you have 15 partitions)
>> >   b) partition replication configuration (each partition necessarily
>> > has
>> a
>> > leader - primary responsible to hold the data - and for reads and
>> > writes)
>> > you can configure the topic to have a number of replicas
>> > 2) How you publish messages to the topic
>> >   a) The publisher is responsible to choose the partition. This can be
>> done
>> > consciously (by setting the partition id while sending the message to
>> > the
>> > topic) or unconsciously (by using the DefaultPartitioner or any other
>> > partitioner scheme).
>> >
>> > All messages sent to a specific partition will be written first to the
>> > leader (meaning that the disk configured for the partition leader will
>> > receive the load) and then replicated to the replica (followers).
>> > Kafka does not automatically distribute the data equally to the
>> > different
>> > brokers - you need to think about your architecture having that in
>> > mind.
>> >
>> > I hope it helps
>> >
>> > On Thu, Aug 6, 2020 at 10:23 PM Péter Nagykátai <st4r.f1sch@gmail.com>
>> > wrote:
>> >
>> > > I initially started with one data disk (mounted solely to hold Kafka
>> > data)
>> > > and recently added a new one.
>> > >
>> > > On Thu, Aug 6, 2020 at 10:13 PM <Manoj.Agrawal2@cognizant.com> wrote:
>> > >
>> > > > What do you mean older disk ?
>> > > >
>> > > > On 8/6/20, 12:05 PM, "Péter Nagykátai" <st4r.f1sch@gmail.com>
>> wrote:
>> > > >
>> > > >     [External]
>> > > >
>> > > >
>> > > >     Yeah, but it doesn't do that. My "older" disks have ~70
>> partitions,
>> > > the
>> > > >     newer ones ~5 partitions. That's why I'm asking what went
>> > > > wrong.
>> > > >
>> > > >     On Thu, Aug 6, 2020 at 8:35 PM <Manoj.Agrawal2@cognizant.com>
>> > wrote:
>> > > >
>> > > >     > Kafka  evenly distributed number of partition on each disk
so
>> in
>> > > > your case
>> > > >     > every disk should have 3/2 topic partitions .
>> > > >     > It is producer job to evenly produce data by partition key
>> > > > to
>> > > topic
>> > > >     > partition .
>> > > >     > How it partition key , it is auto generated or producer
>> > > > sending
>> > key
>> > > > along
>> > > >     > with message .
>> > > >     >
>> > > >     >
>> > > >     > On 8/6/20, 7:29 AM, "Péter Nagykátai" <st4r.f1sch@gmail.com>
>> > > wrote:
>> > > >     >
>> > > >     >     [External]
>> > > >     >
>> > > >     >
>> > > >     >     Hello,
>> > > >     >
>> > > >     >     I have a Kafka cluster with 3 brokers (v2.3.0) and each
>> > broker
>> > > > has 2
>> > > >     > disks
>> > > >     >     attached. I added a new topic (heavyweight) and was
>> surprised
>> > > > that
>> > > >     > even if
>> > > >     >     the topic has 15 partitions, those weren't distributed
>> evenly
>> > > on
>> > > > the
>> > > >     > disks.
>> > > >     >     Thus I got one disk that's almost empty and the other
>> almost
>> > > > filled
>> > > >     > up. Is
>> > > >     >     there any way to have Kafka evenly distribute data on
its
>> > > disks?
>> > > >     >
>> > > >     >     Thank you!
>> > > >     >
>> > > >     >
>> > > >     > This e-mail and any files transmitted with it are for the
>> > > > sole
>> > use
>> > > > of the
>> > > >     > intended recipient(s) and may contain confidential and
>> privileged
>> > > >     > information. If you are not the intended recipient(s), please
>> > reply
>> > > > to the
>> > > >     > sender and destroy all copies of the original message. Any
>> > > > unauthorized
>> > > >     > review, use, disclosure, dissemination, forwarding, printing
>> > > > or
>> > > > copying of
>> > > >     > this email, and/or any action taken in reliance on the
>> > > > contents
>> > of
>> > > > this
>> > > >     > e-mail is strictly prohibited and may be unlawful. Where
>> > permitted
>> > > by
>> > > >     > applicable law, this e-mail and other e-mail communications
>> sent
>> > to
>> > > > and
>> > > >     > from Cognizant e-mail addresses may be monitored.
>> > > >     > This e-mail and any files transmitted with it are for the
>> > > > sole
>> > use
>> > > > of the
>> > > >     > intended recipient(s) and may contain confidential and
>> privileged
>> > > >     > information. If you are not the intended recipient(s), please
>> > reply
>> > > > to the
>> > > >     > sender and destroy all copies of the original message. Any
>> > > > unauthorized
>> > > >     > review, use, disclosure, dissemination, forwarding, printing
>> > > > or
>> > > > copying of
>> > > >     > this email, and/or any action taken in reliance on the
>> > > > contents
>> > of
>> > > > this
>> > > >     > e-mail is strictly prohibited and may be unlawful. Where
>> > permitted
>> > > by
>> > > >     > applicable law, this e-mail and other e-mail communications
>> sent
>> > to
>> > > > and
>> > > >     > from Cognizant e-mail addresses may be monitored.
>> > > >     >
>> > > >
>> > > >
>> > > > This e-mail and any files transmitted with it are for the sole use
>> > > > of
>> > the
>> > > > intended recipient(s) and may contain confidential and privileged
>> > > > information. If you are not the intended recipient(s), please reply
>> to
>> > > the
>> > > > sender and destroy all copies of the original message. Any
>> unauthorized
>> > > > review, use, disclosure, dissemination, forwarding, printing or
>> copying
>> > > of
>> > > > this email, and/or any action taken in reliance on the contents of
>> this
>> > > > e-mail is strictly prohibited and may be unlawful. Where permitted
>> > > > by
>> > > > applicable law, this e-mail and other e-mail communications sent to
>> and
>> > > > from Cognizant e-mail addresses may be monitored.
>> > > > This e-mail and any files transmitted with it are for the sole use
>> > > > of
>> > the
>> > > > intended recipient(s) and may contain confidential and privileged
>> > > > information. If you are not the intended recipient(s), please reply
>> to
>> > > the
>> > > > sender and destroy all copies of the original message. Any
>> unauthorized
>> > > > review, use, disclosure, dissemination, forwarding, printing or
>> copying
>> > > of
>> > > > this email, and/or any action taken in reliance on the contents of
>> this
>> > > > e-mail is strictly prohibited and may be unlawful. Where permitted
>> > > > by
>> > > > applicable law, this e-mail and other e-mail communications sent to
>> and
>> > > > from Cognizant e-mail addresses may be monitored.
>> > > >
>> > >
>> >
>>
>


-- 



*William Reynolds**Technical Operations Engineer*


<https://www.facebook.com/instaclustr>   <https://twitter.com/instaclustr>
<https://www.linkedin.com/company/instaclustr>

Read our latest technical blog posts here
<https://www.instaclustr.com/blog/>.

This email has been sent on behalf of Instaclustr Pty. Limited (Australia)
and Instaclustr Inc (USA).

This email and any attachments may contain confidential and legally
privileged information.  If you are not the intended recipient, do not copy
or disclose its content, but please reply to this email immediately and
highlight the error to the sender and then immediately delete the message.

Instaclustr values your privacy. Our privacy policy can be found at
https://www.instaclustr.com/company/policies/privacy-policy

Mime
View raw message