No More Correlizer Beta!

[sangyup81]

Seems they've come out with Correlizer Applications 1.07 which is the same app.

http://svahesrv2.bioquant.uni-heidelberg.de/correlizer/forum_thread.php?id=48

"The 'Correlizer Applications' are the tested 'Correlizer Beta Applications' and should be used instead of the 'BioMedical Genome Correlations' (when possible). Job submission for the Beta and BioMedical Genome Correlations group has been disabled and unsent jobs will be removed. Although since this is done from the bridge side there will be cases where some of those unsent jobs are submitted, due to the time it takes to remove them. We will look how to grant the credits in case there are systems that complete those jobs before they are removed.

Anis Abuseiris
Erasmus Grid Office"

 

[Fullmetal Chocobo]

So should be abort all existing Correlizer Beta tasks?

 

[Rudy Toody]

Be sure to go into the Correlizer preferences and change your choice to Correlizer applications. Uncheck beta, embedded, and Bio.

 

[waffleironhead]

So should be abort all existing Correlizer Beta tasks?

i'd finish any you have. this will help them clean up better.

 

[wayliff]

Be sure to go into the Correlizer preferences and change your choice to Correlizer applications. Uncheck beta, embedded, and Bio.

done - thanks to the OP for the heads up.

 

[Uppsala9496]

1.07 is running a lot faster than 1.0. It's about 5 minutes faster on my X6 running Win7x64.

 

[Sunny129]

i'd finish any you have. this will help them clean up better.

hmm...except that it'll result in several invalids and/or errors for the user who crunch through all the v1.00 tasks in his or her queue. as a matter of fact, i didn't have any errors or invalids yesterday, but when i woke up this morning i had 19 errors and 22 invalids. the 19 errors were all v1.00 tasks that were canceled server side (according to the error message on the server) before they had a chance to start crunching on my machine...mind you these tasks were all downloaded well before the decision to stop sending out v1.00 tasks. the 22 invalids were all v1.00 tasks that ran to completion, but got no credit b/c again, they were canceled server side (again, according to the server error message). the OP, sangyup81, was my wingman on many of those invalids. of those tasks for which we were wingmen, my task status reads "completed, can't validate" for all of them, while his task status reads "aborted by user" for all of them. his abortions all took place at Nov. 2, 19:00:40 UTC, and occurred before any of those tasks started crunching (i.e. he wasted no CPU cycles on those tasks)...i on the other hand wasted 22 tasks @ 980s = 21,560s = ~6 hours of CPU time.

i'm still not sure if a few, a lot, or all of my remaining v1.00 tasks will end up this way, and so i'll continue crunching through them for now. but if anyone has a second opinion on whether we should abort any remaining v1.00 tasks in our queues, i'm all ears...

 

[waffleironhead]

Sunny129

It looks to me like they stopped producing any aditional WU's for the 1.00 client and then error/invalidate out any tasks which had no chance of being validated, since they werent going to send out any more wu's.
It kinda sucks you spent the time cpu time on those tasks with no chance of validation, but the admins have been good in the past on making things right.

If anyone decides to abort the 1.00 WU's, you will be abandoning your wingman. For those of you whose wingman have abandoned you, we shall hope the admins make things right.

I'll have to keep an eye on my account and watch the correlizer forums.

 

[Sunny129]

Sunny129

It looks to me like they stopped producing any aditional WU's for the 1.00 client and then error/invalidate out any tasks which had no chance of being validated, since they werent going to send out any more wu's.

ok, i get it now...the minimum quorum per task is 2, and so the initial replication is also 2. now, b/c my wingman sangyup81 aborted his v1.00 tasks before the project even announced publicly in the news section of the Correlizer forum that v1.00 tasks would be phased out of production, 1 of the 2 initial replications of all of those tasks have been invalidated already. a 3rd task replication would have to be sent out to another host and completed in order to achieve the minimum quorum of 2 and confirm validation, but b/c of the phase-out of v1.00 tasks, no 3rd replication will get sent out. therefore, my completed task never stands a chance of getting validated, and thus gets marked invalid.

darn you sangyup81 for aborting all those tasks ...we would have both returned usable scientific results (and gotten credit) for our remaining v1.00 tasks had you not aborted...i'd go through my list of v1.00 "tasks in progress" to see if any other wingmen have aborted, but the list is several hundred tasks long, and the data driven web pages of the Correlizer server load at a snail's pace...so i guess i'll just crunch through them all and accept that some of them (possibly a great deal of them) will error out/get marked invalid.

It kinda sucks you spent the time cpu time on those tasks with no chance of validation, but the admins have been good in the past on making things right.

i'm not exactly sure what you mean by that, but if you're referring to credit that should have been earned but wasn't, i don't care too much for credit. don't get me wrong, i know that in most situations credit is more than just credit - its a direct indication of a host's scientific productivity. but in this instance, the CPU cycles that got wasted i can never get back. even if they gave me credit for those wasted CPU cycles, i still know in the back of my head that they did nothing scientifically productive b/c the project never ended up using that crunched data. so if they decide to award credit for that, great. if not, that's fine with me too. i don't care if i don't get credit for work done when that work wasn't used by the project anyways...

 

[sangyup81]

sorry Sunny129, I had no idea they were going to abandon the 1.00 tasks. I aborted those 1.00 tasks because they went to machines that were supposed to run betas. when betas became available again, I aborted the 1.00 tasks. =X

I still have some 1.00 tasks running in my other machines. Should I just abort those too?

 

[wayliff]

As for me, I decided not to abort anything. For what I see it should process those WUs within several hours from now.
I do have a couple of machines that may take longer processing those last WUs.

 

[Sunny129]

sorry Sunny129, I had no idea they were going to abandon the 1.00 tasks. I aborted those 1.00 tasks because they went to machines that were supposed to run betas. when betas became available again, I aborted the 1.00 tasks. =X

I still have some 1.00 tasks running in my other machines. Should I just abort those too?

NO...don't abort them unless you look at each individual task first and make sure that your wingman has already aborted the task. if your wingman hasn't aborted the task yet, do not abort yours - if both finish without error, it'll meet the minimum quorum of 2, and both of yours and your wingman's tasks should get validated and earn credit. i still have hundreds of v1.00 tasks left in my queue, and b/c the pages that display task information on my account have been loading dog slow lately, its impossible for me to check each task without being in front of the computer for the next 24 hours lol...so i'm just letting them all run to completion.

 

[sangyup81]

oh man, you're right, this is going to take a while lol i'll just abort the new 1.07 apps on those machines and then set my additional work buffer to 0 =P

the only reason I don't have it to 0 is because of Milkyway's problems lately but I think they're resolved now

 

[Sunny129]

oh man, you're right, this is going to take a while lol i'll just abort the new 1.07 apps on those machines and then set my additional work buffer to 0 =P

the only reason I don't have it to 0 is because of Milkyway's problems lately but I think they're resolved now

why do you need to abort the 1.07's? why don't you just let them run to completion too? after all, we do need all the credit we can get if we intend to win this race...

 

[sangyup81]

The non-beta 1.07 apps will be reassigned so they can be aborted safely. This way, I can get the old apps done faster and with the 1.07s reassigned, they will get validated faster vs. me suspending them to give the old apps preference. As for the betas and the 1.00s, I'll be aborting them selectively

 

[VirtualLarry]

Huh. My main machine, is running only Correlizer 1.07 tasks. My other machine, is still working through a pile of the BioMedical 1.00 tasks. I'm going to let them finish processing, unless someone tells me otherwise.

 

[sangyup81]

I think the worry is that if you're crunching a 1.00 and your wingman has aborted, you might not get credit. VirtualLarry, I think this is only a big problem if you have a lot of the 1.00's remaining. If it's less than a day's worth, I wouldn't sweat it.

 

[GLeeM]

it'll meet the minimum quorum of 2, and both of yours and your wingman's tasks should get validated and earn credit

Many or most of my "pending" are showing both me and my wingman with this text:
Completed, waiting for validation

Shouldn't these be "valid"?

 

[brownstone]

Am I doing the same correlizer as you guys? Where is all this wingman info at and where do I find all this glorious information? I guess I've just been chugging along with my head in the sand!

 

[Uppsala9496]

On the correlizer site you can click on the tasks for your computer and from there it tells you who your wingman is.
The site is horribly slow though.

 

[Sunny129]

Many or most of my "pending" are showing both me and my wingman with this text:
Completed, waiting for validation

Shouldn't these be "valid"?

yes, i would think so...thanks for bringing that to our attention. i just kinda assumed that my pending v1.00 tasks were just waiting on wingmen, but now i know that some pending tasks aren't waiting for wingmen and still aren't validated. so i'll have to check to see if some of my pending v1.00 tasks are ending up in the same state.

on the other hand, many of my pending v1.00 tasks continue to get validated. just to give you an idea, ~6 hours ago i had 345 pending, 842 valid, 22 invalid, and 19 errors. currently i have 459 pending, 883 valid, 22 invalid, and 19 errors...and yes, these are all v1.00 tasks, the most recent of which was validated 4 Nov. 2011, 12:54:38 UTC. just wanted to show that over the past 6 hours or so, some of my pending v1.00 tasks have continued to validate, and i haven't picked up any invalids or errors.

 

[brownstone]

On the correlizer site you can click on the tasks for your computer and from there it tells you who your wingman is.
The site is horribly slow though.

Thanks for the response Uppsala9496!

 

[bryanW1995]

Be sure to go into the Correlizer preferences and change your choice to Correlizer applications. Uncheck beta, embedded, and Bio.

Done.

 

[bryanW1995]

hmm...except that it'll result in several invalids and/or errors for the user who crunch through all the v1.00 tasks in his or her queue. as a matter of fact, i didn't have any errors or invalids yesterday, but when i woke up this morning i had 19 errors and 22 invalids. the 19 errors were all v1.00 tasks that were canceled server side (according to the error message on the server) before they had a chance to start crunching on my machine...mind you these tasks were all downloaded well before the decision to stop sending out v1.00 tasks. the 22 invalids were all v1.00 tasks that ran to completion, but got no credit b/c again, they were canceled server side (again, according to the server error message). the OP, sangyup81, was my wingman on many of those invalids. of those tasks for which we were wingmen, my task status reads "completed, can't validate" for all of them, while his task status reads "aborted by user" for all of them. his abortions all took place at Nov. 2, 19:00:40 UTC, and occurred before any of those tasks started crunching (i.e. he wasted no CPU cycles on those tasks)...i on the other hand wasted 22 tasks @ 980s = 21,560s = ~6 hours of CPU time.

i'm still not sure if a few, a lot, or all of my remaining v1.00 tasks will end up this way, and so i'll continue crunching through them for now. but if anyone has a second opinion on whether we should abort any remaining v1.00 tasks in our queues, i'm all ears...

I just went through and aborted all of my 1.0 tasks just to be safe.

 

[sangyup81]

I decided to do it the hard way and suspend all 1.00 tasks that aren't finished yet by my wingmen

Once all my unsuspended tasks are done in a few days, I'll take a look at those tasks again