Suddenly, Most Tasks Marked "Validate Error"?
5 Apr 2022 17:57:35 UTC
Topic 227336
(moderation:
Today I found many "Validation Error" tasks. I'm not sure I've seen this happen before. Seems like it's happening to all my work all of a sudden. It's hard to tell if it's all tasks or just some tasks. I think I've received credit for some recent tasks, but not the most recent tasks. Also, it looks like E@H is trying to re-validate by created new tasks for the same WU and sending them out. I guess that makes sense, but it's not usually necessary, right? Even when my tasks are judged invalid, it only takes three results, not four or five. Something is different.
Can someone explain?
Here is an example WU: https://einsteinathome.org/workunit/623832532
you have 33 tasks marked as
)
you have 33 tasks marked as invalid due to the validation errors, but over 1600 valid tasks. I wouldn't call that "most".
maybe an issue with the project validator on these tasks and they had to be resent.
i wouldnt worry about it. you have hundreds of valid tasks from just the past 2 days.
_________________________________________________________________________
I also have received many
)
I also have received many (nearly 60 so far) tasks of "Validation Error". Of importance to me is the statement where it says: "Couldn't open file "LATeah3012L05....." No such file or directory"
Here is a shot of what I'm talking about.
I would like at least some explanation of what this means, and why we're all of a sudden getting it.
Proud member of the Old Farts Association
GWGeorge007 wrote: I also
)
the stderr output doesn't provide anything useful for this. thats just information about the task itself (from your system). it shows the task completed fine.
the validation happens on the project's servers and we don't have any insight into what it's doing unless a project admin posts about it.
_________________________________________________________________________
Ian, you might want to look
)
Ian, you might want to look at your own computer's "Validation Error", computer #12803503. You, too, have many such tasks.
Could you contact Bernd and see what's up?
.....[EDIT[.....
You seem to have a better connection with him than I do.
Proud member of the Old Farts Association
yes I see them. everyone
)
yes I see them. everyone likely has them, I have more than most because my system(s) are processing many more tasks than most. which means it's not a problem with anyone's specific computer and something wrong with the project validator. there's nothing you can or need to do about it. if the project has an issue with the validator, they will see it and fix it when able.
I sent Bernd a message earlier (about something else, but also project server related) but haven't heard back. He's busy and doesnt always answer.
my hunch is that there was a problem with the validator not validating a bunch of results. maybe they were stuck. they unstuck it, but it errored out all the stuck ones, which needed to be sent back out. just my guess.
edit, just double checked. looks like most validation has quit around 1500 UTC today. the admins will need to fix it. but they usually award credits retro actively for problems like this. I'll try to cross post the issue into one of the technical forums that Bernd frequents.
_________________________________________________________________________
I as well have had many, many
)
I as well have had many, many more than what I would usually get... if any at all in the past.
There must be something different with this new app for NVIDIA GPUs causing this issue. But that doesn't appear to be the case with PAUL as he has AMD GPU.
I just started running E@H again about 5 days ago and these are my Invalid totals:
1x 3080ti machine = 209 Invalids and counting
1x 3070ti machine = 144 Invalids and counting
1x 3060ti machine = 104 Invalids and counting (only been 3 days)
2x 1660ti machine = 75 Invalids and counting (only been 2 days)
That is a lot of points gone to the crapper and quite a bit of wasted resources IMO. Like about 10-15% minimum Invalid rate.
Same here. Nvidia 750Ti(s).
)
Same here. Nvidia 750Ti(s). Dozen for 5 of Apr
bluestang wrote: I as well
)
has nothing to do with the nvidia app. it's affecting everyone just the same no matter the system or application (for FGRPB1G).
I've seen the admins award credits retroactively for problems like this that aren't the user's fault. wouldnt be surprised if they do it here too.
or you could suspend Einstein and crunch something else until they fix it.
_________________________________________________________________________
Wonder if the recent change
)
Wonder if the recent change to the validators for BRP4 ARM64 tasks and new 1.61 application is connected somehow.
Thanks everyone. I
)
Thanks everyone. I appreciate your responses. I feel better now about waiting. Was worried something on my end was broken.