[jboss-jira] [JBoss JIRA] Commented: (JGRP-560) Message greater than 16286 bytes is not received on MaxOS

Mon Oct 8 07:11:13 EDT 2007

    [ http://jira.jboss.com/jira/browse/JGRP-560?page=comments#action_12380916 ] 

Manik Surtani commented on JGRP-560:
------------------------------------

While Jason's fix above does work, keep in mind that *some* OS X updates reset MTU settings for lo0!  Something I just noticed.  

> Message greater than 16286 bytes is not received on MaxOS
> ---------------------------------------------------------
>
>                 Key: JGRP-560
>                 URL: http://jira.jboss.com/jira/browse/JGRP-560
>             Project: JGroups
>          Issue Type: Bug
>            Reporter: Bela Ban
>         Assigned To: Bela Ban
>             Fix For: 2.6
>
>
> I'm new to JGroups, and I have a Mac OS X (10.4)-specific question. I've used
> JGroups 2.4.1sp3, 2.5.0CR1, and the new 2.5.0GA - I see the same issue with
> all versions.
> I've noticed that, for UDP, there seems to be a max message size of 16286
> bytes (a little less than 16K). This is using the default udp.xml config,
> and just a simple message to another node like:
> JChannel ch = new JChannel("udp.xml");
> ch.setReceiver(this);
> ch.connect("TEST");		
> //
> // 'target' is another group member besides myself
> //
> byte[] buf = new byte[size];
> Arrays.fill(buf, (byte)77);
> echo("Sending message: " + size + " bytes");
> Message msg = new Message(target, null, buf);
> ch.send(msg);
> A receiver node will NOT receive the message if 'size' is greater than
> 16286. In fact, after sending a too-large message, the receiver sees no
> message trafffic so it starts doing heartbeats and view changes). I've
> attached some debug output below.
> TCP transport does not have this problem.
> I'm aware of the tos="8" attribute warning in OS X, and changing the value
> to 0 to avoid the warning doesn't affect the max size issue. Though perhaps
> it's the same Mac quirk causing both problems...
> I've searched around, but haven't seen anything about this. Is it a known
> issue? Is there a workaround? Am I doing something wrong?
> Thanks in advance,
> Rick Pike
> Debug output:
> -------------------------
> SENDER:
> (Sends increasing size messages, 2 sec apart. As soon as the size goes
> beyond 16286, it notices a problem).
> Sending message: 16284 bytes
> Sending message: 16285 bytes
> Sending message: 16286 bytes
> 2007-07-17 12:06:25,934 [DEBUG] FD sending are-you-alive msg to
> 192.168.2.71:51607 (own address=192.168.2.71:51610)
> 2007-07-17 12:06:25,936 [DEBUG] FD received ack from 192.168.2.71:51607
> Sending message: 16287 bytes
> Sending message: 16288 bytes
> RECEIVER:
> (Gets messages OK until 16287. I assume FD starts because it sees no message
> traffic.)
> Received message: 16284 bytes
> Received message: 16285 bytes
> Received message: 16286 bytes
> 2007-07-17 12:06:25,876 [DEBUG] FD sending are-you-alive msg to
> 192.168.2.71:51610 (own address=192.168.2.71:51607)
> 2007-07-17 12:06:25,884 [DEBUG] FD received ack from 192.168.2.71:51610
> 2007-07-17 12:06:29,305 [DEBUG] MERGE2
> initial_mbrs=[[own_addr=192.168.2.71:51610, coord_addr=192.168.2.71:51607,
> is_server=true], [own_addr=192.168.2.71:51607,
> coord_addr=192.168.2.71:51607, is_server=true]]
> 2007-07-17 12:06:35,877 [DEBUG] FD sending are-you-alive msg to
> 192.168.2.71:51610 (own address=192.168.2.71:51607)
> 2007-07-17 12:06:35,880 [DEBUG] FD received ack from 192.168.2.71:51610
> 2007-07-17 12:06:42,496 [DEBUG] MERGE2
> initial_mbrs=[[own_addr=192.168.2.71:51610, coord_addr=192.168.2.71:51607,
> is_server=true], [own_addr=192.168.2.71:51607,
> coord_addr=192.168.2.71:51607, is_server=true]]
> 2007-07-17 12:06:45,879 [DEBUG] FD sending are-you-alive msg to
> 192.168.2.71:51610 (own address=192.168.2.71:51607)
> 2007-07-17 12:06:45,880 [DEBUG] FD received ack from 192.168.2.71:51610
> 2007-07-17 12:06:50,453 [DEBUG] FD_SOCK peer closed socket normally
> 2007-07-17 12:06:50,453 [DEBUG] FD_SOCK pinger thread terminated

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira