[
https://issues.apache.org/jira/browse/TS-3121?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14167551#comment-14167551
]
Sudheer Vinukonda commented on TS-3121:
---------------------------------------
A simple fix for this issue -
{code}
diff --git a/proxy/FetchSM.cc b/proxy/FetchSM.cc
index 596e9fb..8188dd7 100644
--- a/proxy/FetchSM.cc
+++ b/proxy/FetchSM.cc
@@ -245,8 +245,13 @@ FetchSM::InvokePluginExt(int fetch_event)
}
if (!has_sent_header) {
- contp->handleEvent(TS_FETCH_EVENT_EXT_HEAD_DONE, this);
- has_sent_header = true;
+ if (fetch_event != TS_EVENT_VCONN_EOS) {
+ contp->handleEvent(TS_FETCH_EVENT_EXT_HEAD_DONE, this);
+ has_sent_header = true;
+ } else {
+ contp->handleEvent(fetch_event, this);
+ goto out;
+ }
}
// TS-3112: always check 'contp' after handleEvent()
{code}
> Seeing garbage Spdy responses sometimes
> ---------------------------------------
>
> Key: TS-3121
> URL: https://issues.apache.org/jira/browse/TS-3121
> Project: Traffic Server
> Issue Type: Bug
> Components: SPDY
> Affects Versions: 5.0.1
> Reporter: Sudheer Vinukonda
> Assignee: Sudheer Vinukonda
> Fix For: 5.2.0
>
>
> Seeing some HTTP/0.9 version garbage responses with status=0 from SPDY
> sometimes. In some error cases, when http layer cleans up itself and sends a
> FETCH ERROR to the SPDY layer and SPDY layer seems to build a SYN_REPLY with
> garbage HTTP headers. Adding a simple protection to prevent this.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
