OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

ws-tx message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Re: [ws-tx] Issue 092 - WS-BA: specify 'presume compensate' assumption


I think there may be a middle road here. "Record everything" is too 
heavy. I think Ram is right that heuristic reporting involves additional 
logging. This is one of the aspects of interoposition that may require 
additional specification, also.

Alastair

Mark Little wrote:
>
> On 28 Sep 2006, at 02:37, Ram Jeyaraman wrote:
>
>> Presume compensate assumption has some inherent problems as described 
>> below:
>>
>> App1 sends a DO message to App2. Coordinator (App1 site) decides to 
>> forget (presume compensate since no vote has been recorded). The 
>> participant (App2 site) times out and takes the presume-compensate 
>> route. But it hits a snag, and sends Fail. Coordinator receives Fail, 
>> but does not remember the activity anymore; so it does not propagate 
>> the Fail to its superior. This is a problem.
>
> How is this any different to a traditional participant "hitting a 
> snag" if it decides to roll back in 2PC with presumed abort in place? 
> Ultimately there are assumptions made through both protocols that, if 
> violated, could seriously mess up the system. We have to assume that 
> users/developers play by the rules and in the cases where they don't 
> (much much less than the 80/20 rule), then that's something 
> implementations can try to remedy if they so wish.
>
> My preference is to go with Tom's proposed resolution at this time.
>
> Mark.
>
>
>
>>
>> Summary: I suggest that we retain the existing presume nothing 
>> assumption as represented by the current state table transitions. 
>> Further, I suggest reversing the resolution to issue 71 so that we 
>> revert to the text: "All state transitions are reliably recorded, 
>> including application state and coordination metadata".
>>
>> -----Original Message-----
>> From: Ram Jeyaraman [mailto:Ram.Jeyaraman@microsoft.com]
>> Sent: Thursday, August 31, 2006 11:45 AM
>> To: Thomas Freund; ws-tx@lists.oasis-open.org
>> Subject: [ws-tx] Issue 092 - WS-BA: specify 'presume compensate' 
>> assumption
>>
>> This issue is identified as 092.
>>
>> Please ensure the subject line "Issue 092 - WS-BA: specify 'presume 
>> compensate' assumption".
>>
>> ________________________________
>> From: Thomas Freund
>> Sent: Wednesday, August 30, 2006 9:18 PM
>> To: ws-tx@lists.oasis-open.org
>> Subject: [ws-tx] NEW Issue - WS-BA: specify 'presume compensate' 
>> assumption
>>
>>
>> Protocol:  WS-BA
>>
>> Artifact:  spec
>>
>> Draft:  BA specification CD 02
>>
>> Link to the document referenced:
>>
>> http://www.oasis-open.org/committees/download.php/18818/wstx-wsba-1.1-spec-cd-02.pdf 
>>
>>
>> Section and PDF line number: see proposed resolution listed below
>>
>> Issue type: design
>>
>> Related issues:
>>
>> Issue Description: WS-BA does not state a 'presume compensate' 
>> assumption
>>
>> Proposed Resolution:
>>
>> After line 73 insert:
>>
>>  *   In the absence of outcome information for a transaction the 
>> transaction is presumed to have compensated.
>>
>> State Table change:
>> The state table (line 520) ParticipantCompletion/Coordinator 
>> View/Inbound Events/:
>>
>> {Completed, Ended} cell should be: (Send Compensate, Ended)
>>
>>  The state table (line 530) CoordinatorCompletion/Coordinator 
>> View/Inbound Events/:
>>
>> {Completed, Ended} cell should be: (Send Compensate, Ended)
>


[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]