[
https://issues.apache.org/jira/browse/HTRACE-159?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14521142#comment-14521142
]
Masatake Iwasaki commented on HTRACE-159:
-----------------------------------------
Some notes on my testing. I think it's ok to address them later in follo-ups.
htraced showed warning message on client disconnection. It could be INFO
message.
{noformat}
2015-04-30T16:29:33+09:00 W: Error reading header bytes: EOF
{noformat}
When I set HTRACED_ADDRESS_KEY to "localhost:9095" (which is port for REST API
and wrong here), it is unclear what is happening from the messages. We could
add a guard or notice in docs for this situation.
{noformat}
15/04/30 16:08:42 INFO util.log: Logging initialized @1373ms
15/04/30 16:08:43 INFO tracing.SpanReceiverHost: Loaded SpanReceiver
HTracedRESTReceiver successfully.
Shutting down htraced receiver for localhost:9095
hrpc_client_call: successfully opened connection
hrpc_client_call: waiting for response
hrpc_client_rcv_resp(::1:9095): error reading response header: 11 (Resource
temporarily unavailable)
htrace_xmit_impl: hrpc_client_call failed.
htraced_xmit(localhost:9095) failed on try 1. Retrying after a delay.
...
{noformat}
> libhtrace.so: use HRPC endpoint of htraced
> ------------------------------------------
>
> Key: HTRACE-159
> URL: https://issues.apache.org/jira/browse/HTRACE-159
> Project: HTrace
> Issue Type: Bug
> Reporter: Colin Patrick McCabe
> Assignee: Colin Patrick McCabe
> Attachments: HTRACE-159.001.patch, HTRACE-159.002.patch
>
>
> libhtrace.so (the htrace C client) should use the HRPC endpoint of htraced.
> This will be more efficient and also avoid unnecessary native dependencies.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
