Re: [Fwd: Mina 2 problem - InvalidMarkException in java.nio.Buffer] - further thoughts

[Yigal Rachman]

Hi, Mark: Thank you for your rapid response - I really appreciate it. Here is a distilled example of the kind of thing I have to do (there are a host of other similar examples).  The Mina application is a network client that communicates with a scientific instrument.  The client must periodically poll the instrument for its data.  I am using a repeating timer to do this.  Whenever the timer expires, it invokes a callback which writes a command to the instrument via Mina; this is where the synchronization is required.  Below is a simplified snippet of code that demonstrates the problem.  I have not sent the actual code because it is wrapped in my own layer of stuff, which just serves to confuse the issue.  Also, I am not showing any synchronization here, although clearly some is needed (the topic of this post). // notes: // this thing manages just one session (the code that enforces this is omitted) // there is a TextLineCodec installed to convert between strings and raw ascii public class DriverProtocolHandler extends IoHandlerAdapter {     // the Timer thread on which all poll timers run     private static final Timer thePollTimerThread = new Timer("thePollTimerThread", true);     // the timer callback task that operates in this instance     private final PollTimerTask pollTimerTask = new PollTimerTask();     // the session for this instance     private IoSession session = null;    // when the session opens     public void sessionOpened(IoSession session) throws Exception {         this.session = session;         // schedule a repeating timer on the given thread (this will repeat until canceled)         long delayMsec = 0;         // start immediately         long periodMsec = 1000;  // each tick is 1000 milliseconds         pollTimerTask.scheduleAtFixedRate(thePollTimerThread, delayMsec, periodMsec);     }     // when the session closes     public void sessionClosed(IoSession session) throws Exception {               // turn out the lights               pollTimerTask.cancel();               this.session = null;     }    // the timer inner class     // [ DurableTimerTask is simply a robust wrapper around java.util.TimerTask ]    //     private class PollTimerTask extends DurableTimerTask {         // this is called once for each tick of the timer         protected void run() throws Exception {             if (session != null) {                 // poll the instrument                 session.write("\n");             }         }     } Please let me know if you need more information. Thanks again, Yigal Mark Webb wrote: I think I understand your problem, but maybe a sample program would make things clearer... On Mon, Feb 25, 2008 at 6:51 PM, Yigal Rachman <[EMAIL PROTECTED]> wrote: Hi, Folks: I am thinking that the synchronizing bug is almost certainly in my application code. Said application uses the Timer class to invoke some asynchronous processing. The methods are all synchronized, but probably on the wrong instance! So my follow-up question is: If I need to ensure that another thread does not conflict with a Mina thread, how do I synchronize it? Yigal Rachman ---------- Forwarded message ---------- From: Yigal Rachman <[EMAIL PROTECTED]> To: dev@mina.apache.org Date: Mon, 25 Feb 2008 14:51:57 -0800 Subject: Mina 2 problem - InvalidMarkException in java.nio.Buffer Hi, Folks: I am biting the bullet and porting to Mina 2.0.0-M1. Thus far, I seem to have resolved all the changes without much pain. I love the look of the revised Mina - elegant and even simpler to use - wow! I have, however, run into a problem which appears to come from the bowels of Mina. Here is the exception I am seeing: java.nio.InvalidMarkException at java.nio.Buffer.reset(Unknown Source) at org.apache.mina.common.AbstractIoBuffer.reset(AbstractIoBuffer.java:321) at org.apache.mina.common.AbstractPollingIoProcessor.writeBuffer(AbstractPollingIoProcessor.java:577) at org.apache.mina.common.AbstractPollingIoProcessor.flushNow(AbstractPollingIoProcessor.java:528) at org.apache.mina.common.AbstractPollingIoProcessor.flush(AbstractPollingIoProcessor.java:469) at org.apache.mina.common.AbstractPollingIoProcessor.access$500(AbstractPollingIoProcessor.java:43) at org.apache.mina.common.AbstractPollingIoProcessor$Worker.run(AbstractPollingIoProcessor.java:681) at org.apache.mina.util.NamePreservingRunnable.run(NamePreservingRunnable.java:51) at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(Unknown Source) at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source) at java.lang.Thread.run(Unknown Source) The message is clear enough: no mark has been set on the buffer, so a reset is not possible. But it does look fishy.. perhaps a thread that is not properly synchronized? Ideas, anyone? Thank you all for your continuing help. Yigal Rachman