thrift-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "James E. King III (JIRA)" <>
Subject [jira] [Commented] (THRIFT-4598) TBinaryProtocol.readMessageBegin() hangs forever in Java
Date Thu, 05 Jul 2018 11:13:00 GMT


James E. King III commented on THRIFT-4598:

Ahh, I get it now - sorry I didn't see this before.  As part of the work on the C++ side I
modified the library so that when you call stop() on the TServer, the implementation can interrupt
ALL the client readers.  My guess is the Java side does not have that feature, which means
the server is essentially held hostage by any connected client.  See THRIFT-3447.  This is
a common problem, also see THRIFT-2696 and THRIFT-3590, THRIFT-4500.  But really, it sounds
like this is now a duplicate of THRIFT-4252 which you previously reported. :)

> TBinaryProtocol.readMessageBegin() hangs forever in Java
> --------------------------------------------------------
>                 Key: THRIFT-4598
>                 URL:
>             Project: Thrift
>          Issue Type: Bug
>          Components: Java - Library
>    Affects Versions: 0.11.0
>         Environment: Ubuntu Linux 16.04 x86_64
>            Reporter: Mario Emmenlauer
>            Assignee: James E. King III
>            Priority: Major
> I run some internal tests for the robustness of our thrift Java server. One of the robustness
tests is to try to connect to the secure server socket with an insecure client and vice versa.
This may seem a slightly diabolic test but it can be a simple user error to forget enabling
or disabling encryption. So I think thrift should handle this gracefully.
> However it seems that this test leads to an infinitely hanging server thread! The thread
blocks the jvm (Java 1.8.0_161 x86_64) from ending. I can see in jstack that the main method
ends and there are almost no threads left, except some garbage collectors and finalizers
plus org.apache.thrift.server.TThreadPoolServer.
> Here is the call stack of the thrift server thread:
> {code:java}
> java.lang.Thread.State: RUNNABLE
> at Method)
> at
> at
> at
> at
> at
> at
> - locked <0x000000067b390fd8> (a
> at
> at org.apache.thrift.transport.TTransport.readAll(
> at org.apache.thrift.protocol.TBinaryProtocol.readAll(
> at org.apache.thrift.protocol.TBinaryProtocol.readI32(
> at org.apache.thrift.protocol.TBinaryProtocol.readMessageBegin(
> at org.apache.thrift.TBaseProcessor.process(
> at org.apache.thrift.server.TThreadPoolServer$
> at java.util.concurrent.ThreadPoolExecutor.runWorker(
> at java.util.concurrent.ThreadPoolExecutor$
> at
> {code}
> I already reduced the server socket timeout to 30sec and the requestTimeout to 10sec.
Is there anything else I can do to timeout readMessageBegin()?

This message was sent by Atlassian JIRA

View raw message