You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafficserver.apache.org by "Sudheer Vinukonda (JIRA)" <ji...@apache.org> on 2015/01/17 04:52:34 UTC
[jira] [Commented] (TS-3306) Keep_alive_out not working
[ https://issues.apache.org/jira/browse/TS-3306?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14281197#comment-14281197 ]
Sudheer Vinukonda commented on TS-3306:
---------------------------------------
{code}
[Jan 16 19:11:38.217] Server {0x7f308e802700} DEBUG: (http_tunnel) [2] [tunnel_run] producer already done
[Jan 16 19:11:38.217] Server {0x7f308e802700} DEBUG: (http_tunnel) [2] producer_handler [http server HTTP_TUNNEL_EVENT_PRECOMPLETE]
[Jan 16 19:11:38.217] Server {0x7f308e802700} DEBUG: (http_redirect) [HttpTunnel::producer_handler] enable_redirection: [0 0 0] event: 2302
[Jan 16 19:11:38.217] Server {0x7f308e802700} DEBUG: (http) [2] [&HttpSM::tunnel_handler_server, HTTP_TUNNEL_EVENT_PRECOMPLETE]
[Jan 16 19:11:38.217] Server {0x7f308e802700} DEBUG: (http_ss) [1] session closing, netvc 0x7f3078015dc0
[Jan 16 19:11:38.217] Server {0x7f308e802700} DEBUG: (http_tunnel) [2] consumer_handler [user agent VC_EVENT_WRITE_COMPLETE]
{code}
>From the logs, it seems to hit the code below, which considers this as a case of chunk truncation.
https://github.com/apache/trafficserver/blob/master/proxy/http/HttpTunnel.cc#L973
{code}
// Everything is already in the buffer so mark the producer as done. We need to notify
// state machine that everything is done. We use a special event to say the producers is
// done but we didn't do anything
{code}
Could be a timing issue with the way origin is sending the body (which in this case is just 1 byte long)
> Keep_alive_out not working
> --------------------------
>
> Key: TS-3306
> URL: https://issues.apache.org/jira/browse/TS-3306
> Project: Traffic Server
> Issue Type: Bug
> Reporter: Thomas Jackson
> Attachments: cop.log
>
>
> I *hope* that I'm just missing something very obvious, but as best I can tell keep_alive_out is not working on 5.0, 5.1, and master. At some point this worked as we vetted the feature when we rolled it out-- I think we were running 4.2 at the time.
> To show the issue I created a test (https://github.com/jacksontj/trafficserver/commit/f913b88666aef6502b8540149a60b2ce25853f9c) which shows the issue.
> In my test case there is a socket server which just returns a 200 OK keep-alive response with the body being the number of HTTP requests this tcp connection has seen. If you run the test you can see that it will always return 1. I thought this could have been some issue timeouts or something, but you can see in the output (below) that ATS immediately closes the connection to origin.
> {code}
> test_basic_proxy (test_keepalive.TestKeepAliveOut) ... connection from ('127.0.0.1', 39298)
> GET / HTTP/1.1
> Host: 127.0.0.1:45807
> Accept-Encoding: gzip
> Accept: */*
> User-Agent: python-requests/2.5.1 CPython/2.6.6 Linux/2.6.32-431.20.3.el6.ipvs.x86_64
> Client-ip: 127.0.0.1
> X-Forwarded-For: 127.0.0.1
> Via: http/1.1 <proxy_name>[2620011950002221C32458757B78BB48] (ApacheTrafficServer/5.3.0)
> sending data back to the client
> Client disconnected
> waiting for a connection
> connection from ('127.0.0.1', 39300)
> GET / HTTP/1.1
> Host: 127.0.0.1:45807
> Accept-Encoding: gzip
> Accept: */*
> User-Agent: python-requests/2.5.1 CPython/2.6.6 Linux/2.6.32-431.20.3.el6.ipvs.x86_64
> Client-ip: 127.0.0.1
> X-Forwarded-For: 127.0.0.1
> Via: http/1.1 <proxy_name>[2620011950002221C32458757B78BB48] (ApacheTrafficServer/5.3.0)
> sending data back to the client
> Client disconnected
> waiting for a connection
> FAIL
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)