[Swift-devel] Q about MolDyn

Ioan Raicu iraicu at cs.uchicago.edu
Mon Aug 6 22:07:00 CDT 2007


Just resending this other email, I don't think it made it through the 
first time...

Ioan Raicu wrote:
> One other thing, in the past, once it got past the first few stages, 
> it would submit about 16500 jobs all at once, and then it would keep 
> sending a few at a time for every few that were completed.... this 
> time, it sent out about 6000 jobs all at once (making the queue go up 
> to 7K+ jobs), but after that, it did not submit any new jobs, despite 
> many jobs completing.... and eventually, the queue went to 0, and it 
> went all idle.... this is very different than what we saw in previous 
> runs!  Whatever happened, it happened in the middle of the experiment, 
> when it only sent the 6K jobs (instead of 16K it would normally send 
> at this stage).  If there is no discrepancy between the # of jobs 
> Swift think it sent Falkon and what Falkon received, then it is beyond 
> me what happened.
>
> Ioan
>
> Veronika Nefedova wrote:
>> Whats up now? Everything has stopped, no errors on swift site...
>> Do you have any errors now?
>>
>> Nika
>>
>> On Aug 6, 2007, at 6:04 PM, Ioan Raicu wrote:
>>
>>> OK, I restarted Falkon as well as there were 12K jobs trying to go 
>>> through, and keeping the entire ANL/UC site busy, although there was 
>>> no Swift on the other end to pick up the notifications...
>>>
>>> here is the new info:
>>>
>>> Falkon Factory Service: 
>>> http://tg-viz-login2:50020/wsrf/services/GenericPortal/core/WS/GPFactoryService 
>>>
>>> Web server: http://tg-viz-login2.uc.teragrid.org:51000/index.htm
>>>
>>> Note that I changed the port #, its now 50020, so don't forget to 
>>> change that before you start Swift...
>>>
>>> Ioan
>>>
>



More information about the Swift-devel mailing list