I seem to be getting quite a few validation errors, does this mean that I will not get any credits for the work done? if not, then I have wasted time and power crunching.:(
Firstly, please be aware that the Cafe is a social area where important technical messages like this one might be overlooked. With this in mind and once you are aware of it, I'll move this thread to the "Problems" board.
Quote:
I seem to be getting quite a few validation errors, does this mean that I will not get any credits for the work done?
Unfortunately, yes, unless one of the project Admins decides to intervene manually. This could happen if it is determined that the problem that has caused this is entirely server side and if there is an available mechanism to fix the problem.
Quote:
if not, then I have wasted time and power crunching.:(
Even if credit was awarded, time and power have been wasted because the same work has been sent to a third host. That can't be reversed. What would be useful is to determine why the events are occurring in the first place and what needs to be done to stop this from continuing.
In each case your computer seems to be successfully completing and uploading the tasks in question. As you have a number of machines registered with the project and to save others going on a hunt to find the problem tasks, this is a link to the results list of the machine which is exhibiting the behaviour. When you actually delve into each result, things appear quite normal from the client perspective.
Validation errors are not an indication of invalid results, so it is quite likely that all the "validation error" tasks listed would have validated perfectly happily (and there would have been no waste of time and power) if the validation process had been able to complete normally. One of the common causes of validation errors is that at the time the validator is called to do its thing, the files to be validated can't actually be found on the server.
A server glitch could cause this but it is puzzling to see results over a lengthy period all suffering the same fate. You will notice that tasks returned up to May 26 were fine. There was then a gap of nearly two days and tasks returned from May 28 onwards are virtually all being listed as "validation error". It doesn't seem likely that there were a series of server glitches over several days that affected just this one host of yours.
Another possible cause is where a host attempts to report results too quickly after uploading them. I believe this happened some time ago where a third party BOINC client was being used to report results immediately and it was discovered that the reporting was being attempted before the uploaded files were physically able to be found by the validator (or something like that).
It would be interesting to know if anything in particular happened with your host during the two day period beteen the last successful tasks and the first validation errors. Did you change anything to do with BOINC during that period?
I've checked the host, and nothing apart fron OS updates have occurred that I can see.
Do you allow the host to run unattended or do you use BOINC Manager to manage things from time to time? Do you run 24/7 or do you shutdown overnight and at weekends? The 2 day gap actually is from Monday to Wednesday - I'd been wondering if it was associated with a weekend but this doesn't appear likely.
When successful results were being returned, (downloaded between May 21-23 and uploaded between May 23-26) the workflow seemed to indicate fairly continuous operation. After the last of the string of successful results, a bunch of work was downloaded on May 26. There were no further uploads until May 28 and they were all validation errors. This pattern of errors continued over all the days up to May 31. There's absolutely no problem per se with bursts of downloads and uploads, but in this case there seems to be an unusually strong link with the validation errors. Can you shed any light on any actions you might have been taking during this period (26-31 May)?
Validation errors
)
Firstly, please be aware that the Cafe is a social area where important technical messages like this one might be overlooked. With this in mind and once you are aware of it, I'll move this thread to the "Problems" board.
Unfortunately, yes, unless one of the project Admins decides to intervene manually. This could happen if it is determined that the problem that has caused this is entirely server side and if there is an available mechanism to fix the problem.
Even if credit was awarded, time and power have been wasted because the same work has been sent to a third host. That can't be reversed. What would be useful is to determine why the events are occurring in the first place and what needs to be done to stop this from continuing.
In each case your computer seems to be successfully completing and uploading the tasks in question. As you have a number of machines registered with the project and to save others going on a hunt to find the problem tasks, this is a link to the results list of the machine which is exhibiting the behaviour. When you actually delve into each result, things appear quite normal from the client perspective.
Validation errors are not an indication of invalid results, so it is quite likely that all the "validation error" tasks listed would have validated perfectly happily (and there would have been no waste of time and power) if the validation process had been able to complete normally. One of the common causes of validation errors is that at the time the validator is called to do its thing, the files to be validated can't actually be found on the server.
A server glitch could cause this but it is puzzling to see results over a lengthy period all suffering the same fate. You will notice that tasks returned up to May 26 were fine. There was then a gap of nearly two days and tasks returned from May 28 onwards are virtually all being listed as "validation error". It doesn't seem likely that there were a series of server glitches over several days that affected just this one host of yours.
Another possible cause is where a host attempts to report results too quickly after uploading them. I believe this happened some time ago where a third party BOINC client was being used to report results immediately and it was discovered that the reporting was being attempted before the uploaded files were physically able to be found by the validator (or something like that).
It would be interesting to know if anything in particular happened with your host during the two day period beteen the last successful tasks and the first validation errors. Did you change anything to do with BOINC during that period?
Cheers,
Gary.
Sorry for posting in the
)
Sorry for posting in the wrong area, and thanks for the reply.
I've checked the host, and nothing apart fron OS updates have occurred that I can see.
RE: I've checked the host,
)
Do you allow the host to run unattended or do you use BOINC Manager to manage things from time to time? Do you run 24/7 or do you shutdown overnight and at weekends? The 2 day gap actually is from Monday to Wednesday - I'd been wondering if it was associated with a weekend but this doesn't appear likely.
When successful results were being returned, (downloaded between May 21-23 and uploaded between May 23-26) the workflow seemed to indicate fairly continuous operation. After the last of the string of successful results, a bunch of work was downloaded on May 26. There were no further uploads until May 28 and they were all validation errors. This pattern of errors continued over all the days up to May 31. There's absolutely no problem per se with bursts of downloads and uploads, but in this case there seems to be an unusually strong link with the validation errors. Can you shed any light on any actions you might have been taking during this period (26-31 May)?
Cheers,
Gary.
I leave to run 24/7 and
)
I leave to run 24/7 and normally let it upload it self, of sometimes click the update button.
Its set for no work units at the moment, but when the wu's are done then I switch it to download more.
The only times its off is after an OS update.