Well, I crunched CEP2 exclusively for the last 5-6 months (sporadically) before switching over to HFCC exclusively the other day. So in my case, there's just two projects involved.

Are you saying that once the actual crunched data is validated and sent to the scientists that the WU reporting data is removed from the site as well? That doesn't make sense to me, because my understanding of the way CEP2 works is that the crunched data is uploaded directly to Harvard servers and is never even on WCG servers. Only the WU results for CEP2 reporting make it to WCG servers (unless I'm mistaken about this).

In any case, my expectation is that I should be able to see WU results history on the WCG site for every WU sent to my machines for some period of time, because the storage requirements for reporting these results is minimal. In the case of CEP2, processing a WU might yield a 30MB file for the scientists - but reporting that result is only takes a few bytes of storage on the WCG website.

Or am I missing something?


Here's a screenshot:


I did quite a bit of CEP2 crunching between last November and March 1st, so seeing this in my results status is, um, confusing to say the least.

Thanks Lawrence, I must've read right over that. Ah, the joys of old age.

But it looks like I've had valid WUs sitting in my results status for more than four days (I'll see if I can confirm that). Is this even possible? Would it be accurate to say that results are listed for a minimum of four days after validation, or do you know for a fact that it's 4 days only (assuming no errors like mine)?

EDIT: I'd also be interested in knowing if the 4 day cap applies to the other status values of Error/No Reply/Invalid/Too Late/Aborted/User Aborted/Server Aborted...

On


Think this will need a little bit more qualification since when I penned that ages ago, as the practise is these weeks with 650,000 - 700,000 results daily to move completed batches much quicker such as for HCC to make space and to keep the scheduler performance up to standards. The longer the live DB, the slower it gets. Anyway think to have observed that HCC work to have moved to the master DB in 24 hours.

Long stuck results often have something in the quorum such as a "No reply", though quorum is complete. The second of the 3 slaggers clearly has that condition, but then if the quorum is complete and there are no ''error'' results either when looking at the quorum detail, it may be that there is still a result(s) out there that never finished, held by the techs or the system due an error. Given that for CEP the data is not on WCG servers, and the daily is 15k-25k opposed to 200k+ for HCC, there is a lesser urge to push this on by hand.

The 94 pages worth is probably DDDT2 or HCMD2. Nico_B's charts display longer past batches MIA. When things start slowing again I'm sure one of the techs will go on a sweeping spree :D

--//--

Not sure what you see, but there's 10 days exact between the 2 dates for the No Reply.

--//--