qpid-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Robbie Gemmell (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (PROTON-1185) [proton-j] writing to the transport stack doesn't flush all bytes to the Network
Date Thu, 28 Apr 2016 18:14:13 GMT

    [ https://issues.apache.org/jira/browse/PROTON-1185?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15262658#comment-15262658
] 

Robbie Gemmell commented on PROTON-1185:
----------------------------------------

Hi Sree,

I'm finishing up for a short vacation, I'll be back during next week, but some notes before
I go MIA:

- In your mail previously you said you had a screenshot showing some details around the issue.
As I mentioned, the mailing list stripped that, so it hasn't been seen yet.
- You mention having traces. Got an example?
- Neither SenderImpl.send(...) or ReceiverImpl.flow(...) actually write bytes into the transport,
its the subsequent 'pending' checks on the transport which drives generating bytes before
returning how many are immediately available for output.
- The above likely makes this a bit different than the other issue, because even if the SSL
wrapper didn't output everything it could at once, unless there was nothing to output it should
simply go out on the next pending() check.
- The above likely means either transport is incorrectly saying there is nothing to output
when there is, or the issue has nothing to do with the transport and the reactor isn't doing
the right thing with the details at some point.

Robbie

> [proton-j] writing to the transport stack doesn't flush all bytes to the Network
> --------------------------------------------------------------------------------
>
>                 Key: PROTON-1185
>                 URL: https://issues.apache.org/jira/browse/PROTON-1185
>             Project: Qpid Proton
>          Issue Type: Bug
>          Components: proton-j
>    Affects Versions: 0.12.2
>         Environment: All
>            Reporter: Sreeram Garlapati
>              Labels: proton-j, send-stuck
>             Fix For: 0.13.0
>
>
> We are experiencing this issue intermittently - writing to Proton-j transport stack doesn't
flush bytes to the Network. The application remains in "stuck state" - until more bytes are
written to that Transport.
> We are experiencing this in both these APIs - SenderImpl.send(...) and ReceiverImpl.flow(...)
- which under-the-covers writes bytes to Transport. We can see our Traces where we are invoking
these APIs - but at this point - the Proton frames (set PN_TRACE_FRM=1) stops.
> We use SSL transport and our Amqp layer is here: https://github.com/Azure/azure-event-hubs/tree/master/java/azure-eventhubs/src/main/java/com/microsoft/azure/servicebus/amqp
> Please let me know what exact information will be useful to diagnose this issue. 
> This issue currently consistently repro's after Sending on 1 Link (on 1 Session - 1 AmqpConnection)
**for a long time**. While I am trying hard to break this down to a simplest possible reproducer
code - I filed the issue to communicate ahead.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@qpid.apache.org
For additional commands, e-mail: dev-help@qpid.apache.org


Mime
View raw message