Re: [jgroups-dev] Random behavior - Not finding peers

Thanks for prompt reply.

1) we can not use TCPPING as we want dynamic discovery using MPING.

I have to write small maven project to simulate the issue.

Configuration:
TCP 
(start_port= 
7899; 
loopback= 
true; 
recv_buf_size= 
20000000; 
send_buf_size= 
640000; 
discard_incompatible_packets= 
true; 
max_bundle_size= 
64000; 
max_bundle_timeout= 
30; 
use_incoming_packet_handler= 
true; 
enable_bundling= 
true; 
use_send_queues= 
false; 
sock_conn_timeout= 
3000; 
skip_suspected_members= 
true; 
use_concurrent_stack= 
true; 
thread_pool.enabled= 
true; 
thread_pool.min_threads= 
1;t 
hread_pool.max_threads= 
25; 
thread_pool.keep_alive_time= 
5000; 
thread_pool.queue_enabled= 
false; 
thread_pool.queue_max_size= 
100; 
thread_pool.rejection_policy= 
run; 
oob_thread_pool.enabled= 
true; 
oob_thread_pool.min_threads= 
1;o 
ob_thread_pool.max_threads= 
8;o 
ob_thread_pool.keep_alive_time= 
5000; 
oob_thread_pool.queue_enabled= 
false; 
oob_thread_pool.queue_max_size= 
100;oob_thread_pool.rejection_policy=run):MPING 
(timeout= 
1000;num_initial_members=3;mcast_addr=228.0.0.5;mcast_port=7753):FD 
(timeout=2000;max_tries=5):VERIFY_SUSPECT(timeout=2000):pbcast.NAKACK 
(gc_lag=100;retransmit_timeout=600,1200,2400,4800):pbcast.STABLE 
(stability_delay=1000;desired_avg_gossip=20000;max_bytes=0):pbcast.GMS 
(print_local_addr=true;join_timeout=5000;shun=true):VIEW_SYNC 
(avg_send_interval=60000)


Regards,
kamal


On Oct 28, 2010, at 4:39 PM, Bela Ban <[email protected]> wrote:

> #1 Maybe (I hope not) 2.6.15.GA was compiled with JDK 6 and you're  
> using JDK 5 to run the code. This can be fixed by either using JDK  
> 6, or checking out the src code and compiling it yourself with JDK 5.
>
> #2 Make sure TCPPING.initial_hosts contains all members of the  
> cluster and/or port_range is >= 2.
>
> If #2 doesn't disappear after the suggested change, post the steps  
> to reproduce it (including configuration)
>
>
> Kamal Dalal wrote:
>> Hi Bela Ban,
>>
>> We found two issues:
>>
>> 1) jgroups-2.6.15.GA.jar has runtime issue: For class  
>> "org.jgroups.Address":
>> bad class file: /.m2/repository/jgroups/jgroups/
>> 2.6.15.GA/jgroups-2.6.15.GA.jar(org/jgroups/Address.class)
>> class file has wrong version 50.0, should be 49.0
>>
>> 2) We tried with 2.6.14GA version, still able to reproduce the  
>> issue 1 out
>> of 3 times.
>>
>> High level information:
>> The project has multiple unix boxes (For simplicity - consider 2  
>> boxes),
>> where java services run.
>> These services are stateful, in terms of only one JVM (out of 2  
>> boxes) will
>> process client requests. While JVM on otherbox is running and  
>> takeover only
>> on first instance not available, detecting through "Jgroup" event.
>> During maitainance cycle, second instance which should start as  
>> "Ready" due
>> to first instance already started, does not able to detect peer.
>> Please, *refer attached image *for graphical information.
>>
>> Note: This is random  behavior but 1 out of 3 times, we see this  
>> issue. High
>> level suspect is, when logs has warning from *
>> ConnectionTable.Connection.Receiver*, we get "initial_mbrs are [] "
>> Restart of faulty instance is the work around for now.
>>
>> Any idea might help us.
>>
>> Thanks in advance,
>> Kamal.
>>
>> On Wed, Oct 27, 2010 at 12:17 PM, Bela Ban <[email protected]> wrote:
>>
>>
>>> 2.6.7 is very old, can you upgrade to the latest 2.6.x release  
>>> (2.6.15) ?
>>>
>>> Once you've done that, if it happens again, I'd need a reproduceable
>>> scenario where this happens
>>>
>>>
>>> Dalal wrote:
>>>
>>>
>>>> JGroup version: 2.6.7 GA
>>>> On startup does not able to find initial members in the cluster,  
>>>> with same
>>>> group name.
>>>> If restart the JVM it finds on second try. The JVM contains  
>>>> multiple
>>>> Groups
>>>> around 10 or so. But happens only to one group and its random to  
>>>> which
>>>> one.
>>>> Any idea ?
>>>>
>>>>
>>>>
>>> --
>>> Bela Ban
>>> Lead JGroups / Clustering Team
>>> JBoss
>>>
>>>
>>>
>>
>>
>> --- 
>> ---------------------------------------------------------------------
>>
>
> -- 
> Bela Ban
> Lead JGroups / Clustering Team
> JBoss
>

------------------------------------------------------------------------------
The Next 800 Companies to Lead America's Growth: New Video Whitepaper
David G. Thomson, author of the best-selling book "Blueprint to a 
Billion" shares his insights and actions to help propel your 
business during the next growth cycle. Listen Now!
http://p.sf.net/sfu/SAP-dev2dev

_______________________________________________
Javagroups-development mailing list