[
https://issues.apache.org/activemq/browse/AMQ-1993?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=47024#action_47024
]
Filip Hanik commented on AMQ-1993:
----------------------------------
{quote}
there is still an opportunity to have a normal IOException interleaved with a
Forced exception.
{quote}
This is not the case. the deregister operation is atomic. deregister will never
call throw twice. However, the oneway method will not stop the normal exception
from propagating the underlying exception to the top. so only one exception
will raise on the invoking thread.
In case of a 'forced timeout exception' there will be two exceptions. On that
is invoked in the onException, there will be a 2nd exception when the async
handler calls socket.close, and terminates the write state.
{quote}
As this is a filter that is added by choice it is not such a big deal but we
may as well iron out the detail. This is a handy feature.
{quote}
not only handy, but essential. It prevents the entire server from a complete
dead lock due to synchronization in the higher layers.
Once that synchronization has been mended, it is still a handy feature, since
it prevents single threads from becoming locked for a very long time.
> Systems hang due to inability to timeout socket write operation
> ---------------------------------------------------------------
>
> Key: AMQ-1993
> URL: https://issues.apache.org/activemq/browse/AMQ-1993
> Project: ActiveMQ
> Issue Type: Bug
> Components: Broker
> Affects Versions: 5.1.0, 5.2.0
> Environment: Unix (Solaris and Linux tested)
> Reporter: Filip Hanik
> Assignee: Gary Tully
> Priority: Critical
> Attachments: patch-1-threadname-filter.patch,
> patch-3-tcp-writetimeout.patch
>
>
> the blocking Java Socket API doesn't have a timeout on socketWrite
> invocations.
> This means, if a TCP session is dropped or terminated without RST or FIN
> packets, the operating system it left to eventually time out the session. On
> the linux kernel this timeout usually takes 15 to 30minutes.
> For this entire period, the AMQ server hangs, and producers and consumers are
> unable to use a topic.
> I have created two patches for this at the page:
> http://www.hanik.com/covalent/amq/index.html
> Let me show a bit more
> ---------------------------------
> "ActiveMQ Transport: tcp:///X.YYY.XXX.ZZZZ:2011" daemon prio=10
> tid=0x0000000055d39000 nid=0xc78 runnable
> [0x00000000447c9000..0x00000000447cac10]
> java.lang.Thread.State: RUNNABLE
> at java.net.SocketOutputStream.socketWrite0(Native Method)
> at java.net.SocketOutputStream.socketWrite(SocketOutputStream.java:92)
> This is a thread stuck in blocking IO, and can be stuck for 30 minutes during
> the kernel TCP retransmission attempts.
> Unfortunately the thread dump is very misleading since the name of the
> thread, is not the destination or even remotely related to the socket it is
> operating on.
> To mend this, a very simple (and configurable) ThreadNameFilter has been
> suggested to the patch, that appends the destination and helps the system
> administrator correctly identify the client that is about to receive data.
> -----------------------------------
> at org.apache.activemq.broker.region.Topic.dispatch(Topic.java:581)
> at org.apache.activemq.broker.region.Topic.doMessageSend(Topic.java:421)
> - locked <0x00002aaaec155818> (a
> org.apache.activemq.broker.region.Topic)
> at org.apache.activemq.broker.region.Topic.send(Topic.java:363)
> The lock being held at this issue unfortunately makes the entire Topic single
> threaded.
> When this lock is being held, no other clients (producers and consumers) can
> publish to/receive from this topic.
> And this lock can hold up to 30 minutes.
> I consider solving this single threaded behavior a 'feature enhancement' that
> should be handled separately from this bug. Because even if it is solved,
> threads still risk being stuck in socketWrite0 for dropped connections that
> still appear to be established.
> For this, I have implemented a socket timeout filter, based on a
> TransportFilter, this filter only times out connections that are actually
> writing data.
> The two patches are at:
> http://www.hanik.com/covalent/amq/patch-1-threadname-filter.patch
> http://www.hanik.com/covalent/amq/patch-3-tcp-writetimeout.patch
> the binary 0000.jar applies to both 5.1 and trunk and can be used today in
> existing environments.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
