[
https://issues.apache.org/jira/browse/CASSANDRA-1743?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12934487#action_12934487
]
T Jake Luciani commented on CASSANDRA-1743:
-------------------------------------------
I don't think it's TFFT it's how we use it :)
if we want it to work for us we need to re-create the transport on each request
otherwise we will have this corruption issue, but re-creating the transports
defeats the purpose of TFastFramed for me.
as it is now the server will create the transport once per thread and reuse it
for all subsequent requests on that thread.
from CustomTThreadPoolService::run()
{code}
inputTransport = inputTransportFactory_.getTransport(client_);
outputTransport = outputTransportFactory_.getTransport(client_);
inputProtocol =
inputProtocolFactory_.getProtocol(inputTransport);
outputProtocol =
outputProtocolFactory_.getProtocol(outputTransport);
while (!stopped_ && processor.process(inputProtocol,
outputProtocol))
{
inputProtocol =
inputProtocolFactory_.getProtocol(inputTransport);
outputProtocol =
outputProtocolFactory_.getProtocol(outputTransport);
}
{code}
> Switch to TFastFramedTransport
> ------------------------------
>
> Key: CASSANDRA-1743
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1743
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Jonathan Ellis
> Fix For: 0.7.0
>
> Attachments: 1743.txt, 1743.txt
>
>
> Forgot that after THRIFT-831 "fast" mode is not the default and is a separate
> transport class.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
