[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2016-05-17 Thread Ismael Juma (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Ismael Juma updated KAFKA-1445:
---
Fix Version/s: (was: 0.10.1.0)
   0.8.2.0

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.8.2.0
>
> Attachments: KAFKA-1445.patch, KAFKA-1445.patch, 
> KAFKA-1445_2014-05-13_11:25:13.patch, KAFKA-1445_2014-05-14_16:24:25.patch, 
> KAFKA-1445_2014-05-14_16:28:06.patch, KAFKA-1445_2014-05-15_15:15:37.patch, 
> KAFKA-1445_2014-05-15_15:19:10.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



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


[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2014-05-16 Thread Guozhang Wang (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Guozhang Wang updated KAFKA-1445:
-

Attachment: KAFKA-1445_2014-05-15_15:19:10.patch

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.9.0
>
> Attachments: KAFKA-1445.patch, KAFKA-1445.patch, 
> KAFKA-1445_2014-05-13_11:25:13.patch, KAFKA-1445_2014-05-14_16:24:25.patch, 
> KAFKA-1445_2014-05-14_16:28:06.patch, KAFKA-1445_2014-05-15_15:15:37.patch, 
> KAFKA-1445_2014-05-15_15:19:10.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2014-05-16 Thread Guozhang Wang (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Guozhang Wang updated KAFKA-1445:
-

Attachment: KAFKA-1445_2014-05-15_15:15:37.patch

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.9.0
>
> Attachments: KAFKA-1445.patch, KAFKA-1445.patch, 
> KAFKA-1445_2014-05-13_11:25:13.patch, KAFKA-1445_2014-05-14_16:24:25.patch, 
> KAFKA-1445_2014-05-14_16:28:06.patch, KAFKA-1445_2014-05-15_15:15:37.patch, 
> KAFKA-1445_2014-05-15_15:19:10.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2014-05-16 Thread Guozhang Wang (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Guozhang Wang updated KAFKA-1445:
-

Attachment: KAFKA-1445_2014-05-14_16:28:06.patch

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.9.0
>
> Attachments: KAFKA-1445.patch, KAFKA-1445.patch, 
> KAFKA-1445_2014-05-13_11:25:13.patch, KAFKA-1445_2014-05-14_16:24:25.patch, 
> KAFKA-1445_2014-05-14_16:28:06.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2014-05-14 Thread Guozhang Wang (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Guozhang Wang updated KAFKA-1445:
-

Attachment: KAFKA-1445_2014-05-14_16:24:25.patch

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.9.0
>
> Attachments: KAFKA-1445.patch, KAFKA-1445.patch, 
> KAFKA-1445_2014-05-13_11:25:13.patch, KAFKA-1445_2014-05-14_16:24:25.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2014-05-14 Thread Guozhang Wang (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Guozhang Wang updated KAFKA-1445:
-

Attachment: KAFKA-1445_2014-05-13_11:25:13.patch

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.9.0
>
> Attachments: KAFKA-1445.patch, KAFKA-1445.patch, 
> KAFKA-1445_2014-05-13_11:25:13.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2014-05-14 Thread Guozhang Wang (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Guozhang Wang updated KAFKA-1445:
-

Attachment: KAFKA-1445.patch

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.9.0
>
> Attachments: KAFKA-1445.patch, KAFKA-1445.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Updated] (KAFKA-1445) New Producer should send all partitions that have non-empty batches when on of them is ready

2014-05-10 Thread Guozhang Wang (JIRA)

 [ 
https://issues.apache.org/jira/browse/KAFKA-1445?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Guozhang Wang updated KAFKA-1445:
-

Attachment: KAFKA-1445.patch

> New Producer should send all partitions that have non-empty batches when on 
> of them is ready
> 
>
> Key: KAFKA-1445
> URL: https://issues.apache.org/jira/browse/KAFKA-1445
> Project: Kafka
>  Issue Type: Bug
>Reporter: Guozhang Wang
> Fix For: 0.9.0
>
> Attachments: KAFKA-1445.patch
>
>
> One difference between the new producer and the old producer is that on the 
> new producer the linger time is per partition, instead of global. Therefore, 
> when the traffic is low, the sender will likely expire partitions one-by-one 
> and send lots of small request containing only a few partitions with a few 
> data, resulting largely increased request rate.
> One solution of it would be to let senders select all partitions that have 
> non-empty batches when on of them is ready.



--
This message was sent by Atlassian JIRA
(v6.2#6252)