On Tue, Jun 28, 2011 at 10:26 AM, Sanjiva Weerawarana <[email protected]>wrote:

> But it is still solved with your fix (wait a bit) right? Yes binary relay
> is designed to not store anything :).


Yes, my fix solved the issue where members join, and we immediately try to
send the message to it. However, while a message is being received, a member
can crash (leave the cluster). In such a case, we cannot failover in all
cases. What we have done is, check the Synapse error code, and then decide
whether we can failover or not. For example, for CONNECT_REFUSED &
CONNECT_TIMEOUT, we can failover since we have not yet started sending the
message out. However, for errors such as CONNECTION_TIMEOUT, SND_ERRO etc.
we cannot failover, since the stream has been already consumed.



>
> Sanjiva.
>
> On Mon, Jun 27, 2011 at 11:30 AM, Afkham Azeez <[email protected]> wrote:
>
>> As per Supun, this is because BinaryRelay consumes the message stream, so
>> there is no way to automatically failover when binary relay is used.
>>
>>
>> On Sun, Jun 26, 2011 at 2:06 AM, Afkham Azeez <[email protected]> wrote:
>>
>>> The root cause of this issue is as follows.
>>>
>>> As soon as a new worker node joins, and its HTTP/S ports become
>>> available, the LB tries to send messages to that node. However, that node
>>> need a few more seconds to "warm up", hence a few initial requests fail. The
>>> failover mechanism kicks in and tries to resends the message to another
>>> instance. However, an error occurs when it tries to retrieve the original
>>> request envelope, and ends up sending an empty message to the worker node,
>>> which shows up as a malformed XML error when the StAX parser tries to parse
>>> the message.
>>>
>>> One measure I have implemented is allowing some warm up time for newly
>>> added worker nodes which solved this problem for the moment. However, the
>>> failover logic is still broken and needs to be fixed.
>>>
>>>
>>> On Sun, Jun 12, 2011 at 9:21 PM, Sanjiva Weerawarana 
>>> <[email protected]>wrote:
>>>
>>>> Not very likely dude :) .. its been in HEAVY use in eBay for a long time
>>>> now. There has to be some other combination of events that triggers this.
>>>>
>>>> Sanjiva.
>>>>
>>>> On Sun, Jun 12, 2011 at 4:25 PM, Sameera Jayasoma <[email protected]>wrote:
>>>>
>>>>> Binary relay impl might be failing to handle messages in high
>>>>> concurrency.
>>>>>
>>>>> Thanks,
>>>>> Sameera
>>>>>
>>>>>
>>>>> On Fri, Jun 10, 2011 at 3:34 PM, Afkham Azeez <[email protected]> wrote:
>>>>>
>>>>>> Folks,
>>>>>> Please note $subject. This happens rarely. We are using binary relay
>>>>>> in the LB.
>>>>>>
>>>>>> --
>>>>>> *Afkham Azeez*
>>>>>> Director of Architecture; WSO2, Inc.; http://wso2.com
>>>>>> Member; Apache Software Foundation; http://www.apache.org/
>>>>>> * <http://www.apache.org/>**
>>>>>> email: **[email protected]* <[email protected]>* cell: +94 77 3320919
>>>>>> blog: **http://blog.afkham.org* <http://blog.afkham.org>*
>>>>>> twitter: 
>>>>>> **http://twitter.com/afkham_azeez*<http://twitter.com/afkham_azeez>
>>>>>> *
>>>>>> linked-in: **http://lk.linkedin.com/in/afkhamazeez*
>>>>>> *
>>>>>> *
>>>>>> *Lean . Enterprise . Middleware*
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Sameera Jayasoma
>>>>> Technical Lead and Product Manager, WSO2 Carbon
>>>>>
>>>>> WSO2, Inc. (http://wso2.com)
>>>>> email: [email protected]
>>>>> blog: http://tech.jayasoma.org
>>>>>
>>>>> Lean . Enterprise . Middleware
>>>>>
>>>>> _______________________________________________
>>>>> Carbon-dev mailing list
>>>>> [email protected]
>>>>> http://mail.wso2.org/cgi-bin/mailman/listinfo/carbon-dev
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Sanjiva Weerawarana, Ph.D.
>>>> Founder, Chairman & CEO; WSO2, Inc.;  http://wso2.com/
>>>> email: [email protected]; phone: +94 11 763 9614; cell: +94 77 787 6880| +1
>>>> 650 265 8311
>>>> blog: http://sanjiva.weerawarana.org/
>>>>
>>>> Lean . Enterprise . Middleware
>>>>
>>>
>>>
>>>
>>> --
>>> *Afkham Azeez*
>>> Director of Architecture; WSO2, Inc.; http://wso2.com
>>> Member; Apache Software Foundation; http://www.apache.org/
>>> * <http://www.apache.org/>**
>>> email: **[email protected]* <[email protected]>* cell: +94 77 3320919
>>> blog: **http://blog.afkham.org* <http://blog.afkham.org>*
>>> twitter: **http://twitter.com/afkham_azeez*<http://twitter.com/afkham_azeez>
>>> *
>>> linked-in: **http://lk.linkedin.com/in/afkhamazeez*
>>> *
>>> *
>>> *Lean . Enterprise . Middleware*
>>>
>>>
>>
>>
>> --
>> *Afkham Azeez*
>> Director of Architecture; WSO2, Inc.; http://wso2.com
>> Member; Apache Software Foundation; http://www.apache.org/
>> * <http://www.apache.org/>**
>> email: **[email protected]* <[email protected]>* cell: +94 77 3320919
>> blog: **http://blog.afkham.org* <http://blog.afkham.org>*
>> twitter: **http://twitter.com/afkham_azeez*<http://twitter.com/afkham_azeez>
>> *
>> linked-in: **http://lk.linkedin.com/in/afkhamazeez*
>> *
>> *
>> *Lean . Enterprise . Middleware*
>>
>>
>
>
> --
> Sanjiva Weerawarana, Ph.D.
> Founder, Chairman & CEO; WSO2, Inc.;  http://wso2.com/
> email: [email protected]; phone: +94 11 763 9614; cell: +94 77 787 6880 | +1
> 650 265 8311
> blog: http://sanjiva.weerawarana.org/
>
> Lean . Enterprise . Middleware
>



-- 
*Afkham Azeez*
Director of Architecture; WSO2, Inc.; http://wso2.com
Member; Apache Software Foundation; http://www.apache.org/
* <http://www.apache.org/>**
email: **[email protected]* <[email protected]>* cell: +94 77 3320919
blog: **http://blog.afkham.org* <http://blog.afkham.org>*
twitter: **http://twitter.com/afkham_azeez*<http://twitter.com/afkham_azeez>
*
linked-in: **http://lk.linkedin.com/in/afkhamazeez*
*
*
*Lean . Enterprise . Middleware*
_______________________________________________
Carbon-dev mailing list
[email protected]
http://mail.wso2.org/cgi-bin/mailman/listinfo/carbon-dev

Reply via email to