Menu
▾
▴
taverna-users
|
From: Paul A. T. <thi...@nc...> - 2008-02-15 21:02:59
|
Hi! I've just downloaded Taverna (1.7) and am trying it for the first time. We're developing some WSDL-based services and I'd like to use Taverna both as a testing platform and for design feedback - e.g. how easy will it be for our users to hook up our services in Taverna or other WSDL-based tools. Many of our services are fairly CPU-intensive, and as such they are queued on a set of back-end servers, rather than giving back "live" results. Our functions will work like: - initial request with input parameters, starts the job running on the server side and returns a "request id" - request id is given to a status check function, which will report e.g. "in progress" or "finished" (or "error", etc.) - once the status is "finished", a result function can be used, with the request id as input, to retrieve the actual results. So my question is, how does this fit in Taverna 1.7, if at all? Is there some way to structure a workflow so that it can do this periodic polling of the status function, while waiting for a process to complete? The FAQ talks about iteration, but I admit I got lost pretty fast, and I'm not sure if or how this applies to my case. Can anyone give me a hint on how to approach this? Do I need to wait for Taverna 2? - Paul -- +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Paul A. | NCBI / NIH Thiessen | thi...@nc... +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |
|
From: Bharathi K. <Bha...@ma...> - 2008-02-18 11:39:33
Attachments:
GridSam_nested.png
|
Hi Paul, I did similar work with tavern 1.x , using nested workflows. Here i adding the workflow diagram. Hope it is helpful. with regards, Bharathi. Paul A. Thiessen wrote: > Hi! I've just downloaded Taverna (1.7) and am trying it for the first > time. We're developing some WSDL-based services and I'd like to use > Taverna both as a testing platform and for design feedback - e.g. how > easy will it be for our users to hook up our services in Taverna or > other WSDL-based tools. > > Many of our services are fairly CPU-intensive, and as such they are > queued on a set of back-end servers, rather than giving back "live" > results. Our functions will work like: > > - initial request with input parameters, starts the job running on the > server side and returns a "request id" > > - request id is given to a status check function, which will report e.g. > "in progress" or "finished" (or "error", etc.) > > - once the status is "finished", a result function can be used, with the > request id as input, to retrieve the actual results. > > > So my question is, how does this fit in Taverna 1.7, if at all? Is there > some way to structure a workflow so that it can do this periodic polling > of the status function, while waiting for a process to complete? The FAQ > talks about iteration, but I admit I got lost pretty fast, and I'm not > sure if or how this applies to my case. > > Can anyone give me a hint on how to approach this? Do I need to wait for > Taverna 2? > > - Paul > > |
|
From: Paul A. T. <thi...@nc...> - 2008-02-18 14:24:40
|
Hi Bharathi! Thanks, that's helpful, at least it gives me a clue where to get started. What I don't understand from your diagram alone is, does the status check (nested workflow) get repeated if the status is not 'done'? That is, how does your 'checkStatus' element work, does it somehow flag the entire nested workflow as failed so that the whole thing retries? Also, I see you have a control link from the nested workflow output to the 'get from url' processor. But doesn't the nested workflow always produce output in 'statusXML', so 'get from url' would always run? I guess I don't understand how a control link works from multiple outputs... - Paul -- +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Paul A. | NCBI / NIH Thiessen | thi...@nc... +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ On Monday, February 18, 2008, 6:00:18 AM, Bharathi Kattamuri wrote: > Hi Paul, > I did similar work with tavern 1.x , using nested workflows. Here i > adding the workflow diagram. Hope it is helpful. > with regards, > Bharathi. > Paul A. Thiessen wrote: >> Hi! I've just downloaded Taverna (1.7) and am trying it for the first >> time. We're developing some WSDL-based services and I'd like to use >> Taverna both as a testing platform and for design feedback - e.g. how >> easy will it be for our users to hook up our services in Taverna or >> other WSDL-based tools. >> >> Many of our services are fairly CPU-intensive, and as such they are >> queued on a set of back-end servers, rather than giving back "live" >> results. Our functions will work like: >> >> - initial request with input parameters, starts the job running on the >> server side and returns a "request id" >> >> - request id is given to a status check function, which will report e.g. >> "in progress" or "finished" (or "error", etc.) >> >> - once the status is "finished", a result function can be used, with the >> request id as input, to retrieve the actual results. >> >> >> So my question is, how does this fit in Taverna 1.7, if at all? Is there >> some way to structure a workflow so that it can do this periodic polling >> of the status function, while waiting for a process to complete? The FAQ >> talks about iteration, but I admit I got lost pretty fast, and I'm not >> sure if or how this applies to my case. >> >> Can anyone give me a hint on how to approach this? Do I need to wait for >> Taverna 2? >> >> - Paul >> >> |
|
From: Bharathi K. <Bha...@ma...> - 2008-02-19 11:34:58
|
Hi Paul, 'CheckStatus' checks the status and if it is not 'done', fails the nested workflow , and retry on failure flag is on, makes the looping, and it containues until status is 'done'. Control will pass to 'web-page-from_url' only when the the nested workflow is executed successfully. with regards, Bharathi. Paul A. Thiessen wrote: > Hi Bharathi! > > Thanks, that's helpful, at least it gives me a clue where to get > started. What I don't understand from your diagram alone is, does the > status check (nested workflow) get repeated if the status is not 'done'? > That is, how does your 'checkStatus' element work, does it somehow flag > the entire nested workflow as failed so that the whole thing retries? > > Also, I see you have a control link from the nested workflow output to > the 'get from url' processor. But doesn't the nested workflow always > produce output in 'statusXML', so 'get from url' would always run? I > guess I don't understand how a control link works from multiple > outputs... > > - Paul > > |
|
From: Paul A. T. <thi...@nc...> - 2008-02-19 13:10:23
|
Hi Bharathi! Thanks again. > 'CheckStatus' checks the status and if it is not 'done', fails the > nested workflow This is the step I'm missing in my own flow - how do you make a single component fail the entire nested workflow? Is it a beanscript command or something? Sorry for all the questions... ;) I appreciate your help. - Paul -- +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Paul A. | NCBI / NIH Thiessen | thi...@nc... +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ On Tuesday, February 19, 2008, 6:31:41 AM, Bharathi Kattamuri wrote: > Hi Paul, > 'CheckStatus' checks the status and if it is not 'done', fails the > nested workflow , and retry on failure flag is on, makes the looping, > and it containues until status is 'done'. Control will pass to > 'web-page-from_url' only when the the nested workflow is executed > successfully. > with regards, > Bharathi. |
|
From: Bharathi K. <Bha...@ma...> - 2008-02-19 13:33:44
|
Hi Paul, It is a beanscript, it fails the workflow by throwing an exception. with regards, bharathi. Bharathi Kattamuri wrote: > > Paul A. Thiessen wrote: >> Hi Bharathi! Thanks again. >> >> >>> 'CheckStatus' checks the status and if it is not 'done', fails the >>> nested workflow >> >> This is the step I'm missing in my own flow - how do you make a single >> component fail the entire nested workflow? Is it a beanscript command or >> something? >> >> Sorry for all the questions... ;) I appreciate your help. >> >> - Paul >> >> > > |
|
From: Paul A. T. <thi...@nc...> - 2008-02-19 14:28:20
|
Hi Bharathi! Bharathi> It is a beanscript, it fails the workflow by throwing an exception. Ok, what type of exception? I'm trying to do simply "throw new Exception();" but it gives me script error messages like "target error: java.lang.Exception". Is there a particular derived exception class used by beanscript? In general, is there documentation anywhere on what one can and can't do with beanscript? I find it very hard to work with, because I don't know what's supported and what isn't... - Paul -- +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Paul A. | NCBI / NIH Thiessen | thi...@nc... +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ On Tuesday, February 19, 2008, 8:27:47 AM, Bharathi Kattamuri wrote: Bharathi> Hi Paul, Bharathi> It is a beanscript, it fails the workflow by throwing an exception. Bharathi> with regards, Bharathi> bharathi. Bharathi> Paul A. Thiessen wrote: >> Hi Bharathi! Thanks again. >> >> >>> 'CheckStatus' checks the status and if it is not 'done', fails the >>> nested workflow >>> >> >> This is the step I'm missing in my own flow - how do you make a single >> component fail the entire nested workflow? Is it a beanscript command or >> something? >> >> Sorry for all the questions... ;) I appreciate your help. >> >> - Paul >> >> |
|
From: Ian D. <ian...@ma...> - 2008-02-19 14:46:47
|
Hello, Try www.beanshell.org . Should have some info on what you can and can't do. Cheers, Ian Ian Dunlop myGrid team School of Computer Science University of Manchester On 19 Feb 2008, at 14:28, Paul A. Thiessen wrote: > Hi Bharathi! > > Bharathi> It is a beanscript, it fails the workflow by throwing an > exception. > > Ok, what type of exception? I'm trying to do simply "throw new > Exception();" but it gives me script error messages like "target > error: > java.lang.Exception". Is there a particular derived exception class > used > by beanscript? > > In general, is there documentation anywhere on what one can and > can't do > with beanscript? I find it very hard to work with, because I don't > know > what's supported and what isn't... > > - Paul > > -- > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ > Paul A. | NCBI / NIH > Thiessen | thi...@nc... > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ > > On Tuesday, February 19, 2008, 8:27:47 AM, Bharathi Kattamuri wrote: > > Bharathi> Hi Paul, > > Bharathi> It is a beanscript, it fails the workflow by throwing an > exception. > > Bharathi> with regards, > Bharathi> bharathi. > Bharathi> Paul A. Thiessen wrote: >>> Hi Bharathi! Thanks again. >>> >>> >>>> 'CheckStatus' checks the status and if it is not 'done', fails >>>> the >>>> nested workflow >>>> >>> >>> This is the step I'm missing in my own flow - how do you make a >>> single >>> component fail the entire nested workflow? Is it a beanscript >>> command or >>> something? >>> >>> Sorry for all the questions... ;) I appreciate your help. >>> >>> - Paul >>> >>> > > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Taverna-users mailing list > Tav...@li... > https://lists.sourceforge.net/lists/listinfo/taverna-users > Documentation: http://www.mygrid.org.uk/usermanual1.6/ > FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq > Biological Services: http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices |
|
From: Paul A. T. <thi...@nc...> - 2008-02-19 15:02:16
Attachments:
exact-search.png
|
Thanks, Ian! Ok, that helps. From reading that, it seems "TargetError"
is beanshell's way of saying that the script threw an exception, which
is reasonable since my script is:
reqid_out = reqid_in;
if (!status_in.equals("eStatus_Success")) {
throw new Exception();
}
So maybe my problem is something else? I can't figure out how to get the
nested workflow to repeat. I've set it e.g. for 10 retries separated by
10000 ms, but when I run the flow (diagram attached), all I get is
that the nested workflow got "ServiceFailure" with the message
Output 'reqid_out' was declared but never created!
... which seems reasonable if the component before threw an exception.
But the whole larger flow just stops there; how to I get the nested
workflow to retry, is there some other flag I have to set explicitly
somewhere?
Sorry to be such a bother with this, I seem to be getting close but am
obviously still missing some subtlety...
- Paul
--
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Paul A. | NCBI / NIH
Thiessen | thi...@nc...
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
On Tuesday, February 19, 2008, 9:49:54 AM, Ian Dunlop wrote:
> Hello,
> Try www.beanshell.org . Should have some info on what you can and
> can't do.
> Cheers,
> Ian
> Ian Dunlop
> myGrid team
> School of Computer Science
> University of Manchester
> On 19 Feb 2008, at 14:28, Paul A. Thiessen wrote:
>> Hi Bharathi!
>>
>> Bharathi> It is a beanscript, it fails the workflow by throwing an
>> exception.
>>
>> Ok, what type of exception? I'm trying to do simply "throw new
>> Exception();" but it gives me script error messages like "target
>> error:
>> java.lang.Exception". Is there a particular derived exception class
>> used
>> by beanscript?
>>
>> In general, is there documentation anywhere on what one can and
>> can't do
>> with beanscript? I find it very hard to work with, because I don't
>> know
>> what's supported and what isn't...
>>
>> - Paul
>>
>> --
>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>> Paul A. | NCBI / NIH
>> Thiessen | thi...@nc...
>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>
>> On Tuesday, February 19, 2008, 8:27:47 AM, Bharathi Kattamuri wrote:
>>
>> Bharathi> Hi Paul,
>>
>> Bharathi> It is a beanscript, it fails the workflow by throwing an
>> exception.
>>
>> Bharathi> with regards,
>> Bharathi> bharathi.
>> Bharathi> Paul A. Thiessen wrote:
>>>> Hi Bharathi! Thanks again.
>>>>
>>>>
>>>>> 'CheckStatus' checks the status and if it is not 'done', fails
>>>>> the
>>>>> nested workflow
>>>>>
>>>>
>>>> This is the step I'm missing in my own flow - how do you make a
>>>> single
>>>> component fail the entire nested workflow? Is it a beanscript
>>>> command or
>>>> something?
>>>>
>>>> Sorry for all the questions... ;) I appreciate your help.
>>>>
>>>> - Paul
>>>>
>>>>
>>
>>
>>
>> -------------------------------------------------------------------------
>> This SF.net email is sponsored by: Microsoft
>> Defy all challenges. Microsoft(R) Visual Studio 2008.
>> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
>> _______________________________________________
>> Taverna-users mailing list
>> Tav...@li...
>> https://lists.sourceforge.net/lists/listinfo/taverna-users
>> Documentation: http://www.mygrid.org.uk/usermanual1.6/
>> FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq
>> Biological Services: http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Taverna-users mailing list
> Tav...@li...
> https://lists.sourceforge.net/lists/listinfo/taverna-users
> Documentation: http://www.mygrid.org.uk/usermanual1.6/
> FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq
> Biological Services:
> http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices |
|
From: Bharathi K. <Bha...@ma...> - 2008-02-19 15:16:30
|
Hi Paul,
set the flag critical =true
you can see in the workflow as
<s:processor name="Nested_Workflow">
<s:workflow maxretries="100" retrydelay="10000" critical="true">
Bharathi.
Paul A. Thiessen wrote:
> Thanks, Ian! Ok, that helps. From reading that, it seems "TargetError"
> is beanshell's way of saying that the script threw an exception, which
> is reasonable since my script is:
>
> reqid_out = reqid_in;
> if (!status_in.equals("eStatus_Success")) {
> throw new Exception();
> }
>
>
> So maybe my problem is something else? I can't figure out how to get the
> nested workflow to repeat. I've set it e.g. for 10 retries separated by
> 10000 ms, but when I run the flow (diagram attached), all I get is
> that the nested workflow got "ServiceFailure" with the message
>
> Output 'reqid_out' was declared but never created!
>
> ... which seems reasonable if the component before threw an exception.
> But the whole larger flow just stops there; how to I get the nested
> workflow to retry, is there some other flag I have to set explicitly
> somewhere?
>
> Sorry to be such a bother with this, I seem to be getting close but am
> obviously still missing some subtlety...
>
> - Paul
>
>
>
> ------------------------------------------------------------------------
>
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> ------------------------------------------------------------------------
>
> _______________________________________________
> Taverna-users mailing list
> Tav...@li...
> https://lists.sourceforge.net/lists/listinfo/taverna-users
> Documentation: http://www.mygrid.org.uk/usermanual1.6/
> FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq
> Biological Services: http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices
|
|
From: Stuart O. <so...@cs...> - 2008-02-19 15:08:51
|
Hi,
I think you need to mark the nested workflow (thats doing the polling)
as critical.
Its a tickbox on the Advanced Model Explorer next to where you set the
number of retries.
Stuart.
On 19 Feb 2008, at 15:02, Paul A. Thiessen wrote:
> Thanks, Ian! Ok, that helps. From reading that, it seems "TargetError"
> is beanshell's way of saying that the script threw an exception, which
> is reasonable since my script is:
>
> reqid_out = reqid_in;
> if (!status_in.equals("eStatus_Success")) {
> throw new Exception();
> }
>
>
> So maybe my problem is something else? I can't figure out how to get
> the
> nested workflow to repeat. I've set it e.g. for 10 retries separated
> by
> 10000 ms, but when I run the flow (diagram attached), all I get is
> that the nested workflow got "ServiceFailure" with the message
>
> Output 'reqid_out' was declared but never created!
>
> ... which seems reasonable if the component before threw an exception.
> But the whole larger flow just stops there; how to I get the nested
> workflow to retry, is there some other flag I have to set explicitly
> somewhere?
>
> Sorry to be such a bother with this, I seem to be getting close but am
> obviously still missing some subtlety...
>
> - Paul
>
> --
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> Paul A. | NCBI / NIH
> Thiessen | thi...@nc...
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>
> On Tuesday, February 19, 2008, 9:49:54 AM, Ian Dunlop wrote:
>
>> Hello,
>
>> Try www.beanshell.org . Should have some info on what you can and
>> can't do.
>
>> Cheers,
>
>> Ian
>
>> Ian Dunlop
>> myGrid team
>> School of Computer Science
>> University of Manchester
>
>
>
>> On 19 Feb 2008, at 14:28, Paul A. Thiessen wrote:
>
>>> Hi Bharathi!
>>>
>>> Bharathi> It is a beanscript, it fails the workflow by throwing an
>>> exception.
>>>
>>> Ok, what type of exception? I'm trying to do simply "throw new
>>> Exception();" but it gives me script error messages like "target
>>> error:
>>> java.lang.Exception". Is there a particular derived exception class
>>> used
>>> by beanscript?
>>>
>>> In general, is there documentation anywhere on what one can and
>>> can't do
>>> with beanscript? I find it very hard to work with, because I don't
>>> know
>>> what's supported and what isn't...
>>>
>>> - Paul
>>>
>>> --
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>> Paul A. | NCBI / NIH
>>> Thiessen | thi...@nc...
>>> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>>>
>>> On Tuesday, February 19, 2008, 8:27:47 AM, Bharathi Kattamuri wrote:
>>>
>>> Bharathi> Hi Paul,
>>>
>>> Bharathi> It is a beanscript, it fails the workflow by throwing an
>>> exception.
>>>
>>> Bharathi> with regards,
>>> Bharathi> bharathi.
>>> Bharathi> Paul A. Thiessen wrote:
>>>>> Hi Bharathi! Thanks again.
>>>>>
>>>>>
>>>>>> 'CheckStatus' checks the status and if it is not 'done', fails
>>>>>> the
>>>>>> nested workflow
>>>>>>
>>>>>
>>>>> This is the step I'm missing in my own flow - how do you make a
>>>>> single
>>>>> component fail the entire nested workflow? Is it a beanscript
>>>>> command or
>>>>> something?
>>>>>
>>>>> Sorry for all the questions... ;) I appreciate your help.
>>>>>
>>>>> - Paul
>>>>>
>>>>>
>>>
>>>
>>>
>>> -------------------------------------------------------------------------
>>> This SF.net email is sponsored by: Microsoft
>>> Defy all challenges. Microsoft(R) Visual Studio 2008.
>>> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
>>> _______________________________________________
>>> Taverna-users mailing list
>>> Tav...@li...
>>> https://lists.sourceforge.net/lists/listinfo/taverna-users
>>> Documentation: http://www.mygrid.org.uk/usermanual1.6/
>>> FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq
>>> Biological Services: http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices
>
>
>> -------------------------------------------------------------------------
>> This SF.net email is sponsored by: Microsoft
>> Defy all challenges. Microsoft(R) Visual Studio 2008.
>> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
>> _______________________________________________
>> Taverna-users mailing list
>> Tav...@li...
>> https://lists.sourceforge.net/lists/listinfo/taverna-users
>> Documentation: http://www.mygrid.org.uk/usermanual1.6/
>> FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq
>> Biological Services:
>> http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices
> <exact-
> search
> .png
> >
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/_______________________________________________
> Taverna-users mailing list
> Tav...@li...
> https://lists.sourceforge.net/lists/listinfo/taverna-users
> Documentation: http://www.mygrid.org.uk/usermanual1.6/
> FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq
> Biological Services: http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices
Stuart Owen, myGrid Team
School of Computer Science
The University of Manchester
|
|
From: Stuart O. <so...@cs...> - 2008-02-19 15:12:41
Attachments:
polling_test.xml
|
Hi again, I've attached an example, that polls once a second for 20 seconds. |
|
From: Paul A. T. <thi...@nc...> - 2008-02-19 15:32:30
|
Hi Stuart (and Bharathi)! THANK YOU! With that example to go by, I finally got mine to work. I think the key was both setting the nested workflow itself as critical (which I tried before, but it wasn't sufficient by itself the way I had things set up), and also setting a component of the nested workflow as critical. I restructured my flow (attached) to match Stuart's model, and now it works. Inside the nested workflow, I set the 'fail_if_false' component as critical; and in the top-level flow, I set the nested workflow as critical. Maybe part of my problem is that in my earlier iteration, the output of the nested flow was not created if the exception was thrown, which made the whole thing halt? Not sure exactly, but, some combination of the (re)structuring of my flow and the critical settings made it work. Again, thanks everyone for your help! - Paul -- +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Paul A. | NCBI / NIH Thiessen | thi...@nc... +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ On Tuesday, February 19, 2008, 10:12:45 AM, Stuart Owen wrote: > Hi again, > I've attached an example, that polls once a second for 20 seconds. |
|
From: Paul A. T. <thi...@nc...> - 2008-02-19 20:10:06
|
Hi all! Now that I've gotten the basic retry polling working, I have a couple of follow-up questions: 1) Say the polling gets an error status where the job is known to have failed - e.g. no further retries should be performed, because the job is dead on the server side. How should I "break out" of the retry loop? By throwing an exception? Or some other more graceful way...? 2) Once I work out a good workflow for this polling status check, I assume there's some way to publish it from our site, so that other Taverna users can simply drop it in as another processor w/o having to reinvent the wheel? Can someone point me to where to get started in setting this up? I guess this would be setting up a new "workflow scaveneger" url... how do I do this on my server? Thanks! - Paul -- +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Paul A. | NCBI / NIH Thiessen | thi...@nc... +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |
|
From: Ian D. <ian...@ma...> - 2008-02-20 11:19:10
|
Hello, For (2) you could publish your workflow to myExperiment (www.myexperiment.org ). It is part of the mygrid platform and is used for sharing workflows between scientists (thinks of it as myspace/facebook for scientists). As you suggested, you can also put your SCUFL file on a webserver, right click on Avaiiable Processors and add New Workflow Scavenger by entering the appropriate URL. (if you are using a Mac it is as easy as enabling web sharing from preferences and then dropping the file in the folder called Sites - otherwise you might have to download Apache and configure). Cheers, Ian Ian Dunlop myGrid team School of Computer Science University of Manchester On 19 Feb 2008, at 20:10, Paul A. Thiessen wrote: > Hi all! Now that I've gotten the basic retry polling working, I have a > couple of follow-up questions: > > 1) Say the polling gets an error status where the job is known to have > failed - e.g. no further retries should be performed, because the > job is > dead on the server side. How should I "break out" of the retry loop? > By > throwing an exception? Or some other more graceful way...? > > 2) Once I work out a good workflow for this polling status check, I > assume there's some way to publish it from our site, so that other > Taverna users can simply drop it in as another processor w/o having to > reinvent the wheel? Can someone point me to where to get started in > setting this up? I guess this would be setting up a new "workflow > scaveneger" url... how do I do this on my server? > > Thanks! > > - Paul > > -- > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ > Paul A. | NCBI / NIH > Thiessen | thi...@nc... > +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ > > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Taverna-users mailing list > Tav...@li... > https://lists.sourceforge.net/lists/listinfo/taverna-users > Documentation: http://www.mygrid.org.uk/usermanual1.6/ > FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq > Biological Services: http://www.mygrid.org.uk/wiki/Mygrid/BiologicalWebServices |
Thanks for helping keep SourceForge clean.
X