kafka-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrey Neporada (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (KAFKA-2063) Bound fetch response size
Date Tue, 02 Aug 2016 09:32:20 GMT

    [ https://issues.apache.org/jira/browse/KAFKA-2063?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15403676#comment-15403676
] 

Andrey Neporada commented on KAFKA-2063:
----------------------------------------

(a) I refer to some new server side setting - something like fetch.partition.max.bytes (?).
Broker setting replica.fetch.max.bytes should be deprecated along with consumer settings fetch.message.max.bytes
and  max.partition.fetch.bytes.

(b) Maybe I am running ahead too much here. In context of this ticket, yes, the only goal
of reordering is to make progress and enforce fairness. And this all can be done on client
side. 

(c) I mean to make fetch request deterministic on server side - fetch responses will go in
order requested by client
(d) Yes, we should clearly document that clients who want to limit entire fetch response should
also deploy some method to avoid starvation/unfairness - either random shuffling or round
robin. Random shuffling seems to be easier to implement and IMHO it will work good enough
for ReplicaFetcherThread.

In general, it looks like most people like to
1) retire partition level limit from fetch request
2) keep fetching order the same as the order of partitions in fetch request

Should I update PR? Any objections?



> Bound fetch response size
> -------------------------
>
>                 Key: KAFKA-2063
>                 URL: https://issues.apache.org/jira/browse/KAFKA-2063
>             Project: Kafka
>          Issue Type: Improvement
>            Reporter: Jay Kreps
>
> Currently the only bound on the fetch response size is max.partition.fetch.bytes * num_partitions.
There are two problems:
> 1. First this bound is often large. You may chose max.partition.fetch.bytes=1MB to enable
messages of up to 1MB. However if you also need to consume 1k partitions this means you may
receive a 1GB response in the worst case!
> 2. The actual memory usage is unpredictable. Partition assignment changes, and you only
actually get the full fetch amount when you are behind and there is a full chunk of data ready.
This means an application that seems to work fine will suddenly OOM when partitions shift
or when the application falls behind.
> We need to decouple the fetch response size from the number of partitions.
> The proposal for doing this would be to add a new field to the fetch request, max_bytes
which would control the maximum data bytes we would include in the response.
> The implementation on the server side would grab data from each partition in the fetch
request until it hit this limit, then send back just the data for the partitions that fit
in the response. The implementation would need to start from a random position in the list
of topics included in the fetch request to ensure that in a case of backlog we fairly balance
between partitions (to avoid first giving just the first partition until that is exhausted,
then the next partition, etc).
> This setting will make the max.partition.fetch.bytes field in the fetch request much
less useful and we  should discuss just getting rid of it.
> I believe this also solves the same thing we were trying to address in KAFKA-598. The
max_bytes setting now becomes the new limit that would need to be compared to max_message
size. This can be much larger--e.g. setting a 50MB max_bytes setting would be okay, whereas
now if you set 50MB you may need to allocate 50MB*num_partitions.
> This will require evolving the fetch request protocol version to add the new field and
we should do a KIP for it.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message