[
https://issues.apache.org/jira/browse/STORM-3364?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16804886#comment-16804886
]
Stig Rohde Døssing commented on STORM-3364:
-------------------------------------------
Here's what I would do:
Check the log as I mentioned above. You should be able to find the log line
where bolt 2 is started (see
https://github.com/apache/storm/blob/v1.1.0/storm-core/src/clj/org/apache/storm/daemon/executor.clj#L840).
Check that it happens before bolt 1 starts emitting.
You may want to enable debug logging for Storm. You do this by calling
setDebug(true) on the Config object you use when submitting the topology. This
will make Storm log every time it transfers a tuple. You should be able to tell
what's happening to the lost tuple then.
Also consider upgrading to the latest Storm version, just in case the issue is
fixed.
> Data is emitted by the first bolt but it is not recieved by the second and
> Storm indicates neither a data loss nor timeout failure.
> -----------------------------------------------------------------------------------------------------------------------------------
>
> Key: STORM-3364
> URL: https://issues.apache.org/jira/browse/STORM-3364
> Project: Apache Storm
> Issue Type: Bug
> Environment: Big data platform
> Reporter: maha
> Priority: Blocker
> Attachments: error_converter.png, error_pred.png
>
>
> Hi Storm community,
> I would like to ask a question about data loss in Storm. My Storm topology
> contains 5 bolts(coded in java):
> The first bolt is a converter that receives the data from the spout and do
> some preliminary transformations(coded also on java), then the second bolt
> receives the emitted converter data, calls a Python script and runs it. In
> this Python code, we added a new function that creates new fields from the
> field received from the first bolt.
> When I submit the topology, +tuples are emitted without any problem from the
> first bolt+ to the second but +some tuples are not received by the second
> bolt+, without any failed information. {color:#205081}*After some time(15-20
> min) it works perfectly*{color}.
> The first bolt log is emitting the values but those values are not received
> by the second bolt (nothing on the logs) without ANY error or any failed ACK.
> *Question:* _*{color:#ff0000}I would like to know why it doesn't work for a
> certain period when we submit the topology (directly) and after a period of
> time those information start being received on the second bolt, all the data
> needed is emitted from the first bolt but I can't find it in the second
> one.{color}*
> The timeout for failed ACK on Storm is 30 sec, Storm indicates neither a data
> loss nor timeout failure.
> Thank you, I will be glad to hear from you.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
