[
https://issues.apache.org/jira/browse/KAFKA-344?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13460604#comment-13460604
]
Jun Rao commented on KAFKA-344:
-------------------------------
Thanks for patch v2. There are still comments in the previous review that are
not addressed:
1.2 Since this is a command line tool, if the program stops because of an
error, we need to print out the exception message and the stacktrace to stdout,
instead of log4j so that users can see what's wrong. log4j output may not
always be in stdout.
1.3. Currently, one can still omit both whitelist and blacklist without being
warned. We need to fix this.
2.1 We shouldn't print the message in the following line since it's binary.
Instead, we should log the size and the topic.
logger.debug("send 08 message: " + bytes.toString());
Good change to force write "serializer.class" to
"kafka.serializer.DefaultEncoder". We shouldn't force write the property
"producer.type" to "sync" though. In most cases, we should probably use async
for better throughput. However, we can let users make that decision. Please
change the comment in the file accordingly.
Also, broker.list in config/producer.properties has the wrong format and
comment. Could you fix it in this patch too?
> migration tool from 0.7 to 0.8
> ------------------------------
>
> Key: KAFKA-344
> URL: https://issues.apache.org/jira/browse/KAFKA-344
> Project: Kafka
> Issue Type: Sub-task
> Components: core
> Affects Versions: 0.8
> Reporter: Jun Rao
> Assignee: Yang Ye
> Priority: Blocker
> Labels: tools
> Fix For: 0.8
>
> Attachments: kafka-0.7.0.jar, kafka_07_producer_performance.patch,
> kafka_344_draft.patch, kafka_344_standalone.diff, kafka_344_v1.diff,
> zkclient-0.1.jar
>
> Original Estimate: 336h
> Remaining Estimate: 336h
>
> Since 0.8 is a non-backward compatible release, we need a migration tool. One
> way to do that is to have a tool that can mirror data from an 0.7 Kafka
> cluster into an 0.8 Kafka cluster. Once the mirror is set up, we can then
> first migrate the consumers to the 0.8 cluster, followed by the producers.
> After that, we can decommission the 0.7 cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
