[
https://issues.apache.org/jira/browse/DISPATCH-1893?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17259974#comment-17259974
]
Ken Giusti commented on DISPATCH-1893:
--------------------------------------
Can reproduce eventually by running 'while ctest -R http1_adaptor"; do sleep 1;
done' ...
On analysis it appears as if the test HTTP1 server is not responding to a
request message sent from the adaptor. Tracing the logs shows:
* that the AMQP message containing the request has been completely received by
the edge router (more != true on last logged frame)
* the server-facing adaptor parser indicates that the entire body of the
request has been sent (Content-Length:393216 for example - non-chunked).
* the HTTP test server (python) is blocked attempting to read 393216 bytes of
content from the socket.
The test then times out, with the test HTTP server blocked reading the body and
the adaptor waiting for the test server to return a response.
One interesting point - after the test times out, the log on the server-facing
router contains:
HTTP_ADAPTOR (trace) [C2] HTTP request/response cancelled!. Octets read: 0
written: *385024* (/home/kgiusti/qpid-d\
ispatch/src/adaptors/http1/http1_server.c:1040)
This is an inconsistency - the router believes that the adaptor has sent the
entire body to the proactor, but the actual count of bytes written to the
proactor is < the content length (393216). This explains why the test server
is blocked on read.
> HTTP1: CI failure (hang) in test_01_concurrent_requests
> -------------------------------------------------------
>
> Key: DISPATCH-1893
> URL: https://issues.apache.org/jira/browse/DISPATCH-1893
> Project: Qpid Dispatch
> Issue Type: Test
> Components: Protocol Adaptors
> Affects Versions: 1.15.0
> Reporter: Ken Giusti
> Assignee: Ken Giusti
> Priority: Critical
> Fix For: 1.15.0
>
>
> Occurs most frequently in Travis CI. Needs investigation.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
