Is there a way to rebuild the repository stats from the completed jobs?
We have an elaborate excel file that we use to report usage metrics to our IT guys, when I paste the stats in to the spread sheet I remove them from the repository.
It appears that the stats are not getting data from EVERY job that finishes. Our farm has been rendering at 100 percent for a week now and when I put the data into my spreadsheet it shows around 30 percent.
I know the first thing that is going to be said is to check my spreadsheet for errors.
I KNOW its not putting everything into the stats because I to a partially completed job and marked the incomplete frames as complete on three random suspended jobs. Only one of those 3 showed up in the statistics.
please help as an upgrade to our farm may ride on our ability to both correct this weird stat behavior as well as rebuild the stats from the completed job list.
thanks
Which version of Deadline are you using?
Also, I checked the stats code, and currently stats for completed jobs are only logged when a job is marked complete by a slave. This means that jobs that are marked complete as a result of marking frames as complete manually don’t make it into the stats. Perhaps this is something we should consider changing.
Cheers,
We are losing a ton of stats when users suspend a job that is 98 percent done because they have all they need or see a problem, then we mark them complete and they vanish
we have deadline 4.0.0.39717
is there a process to generate the stats from the completed jobs?
we have IT waiting on these stats and I dont want to give them to them because my best guess is that 60 percent of our jobs are suspended before completion and therefore never make it into the statistics.
thanks for your help
The stats are separated from the jobs, and unfortunately, there isn’t a way to force Deadline to collect the stats from the current completed jobs in the queue. I think we need to modify the system so that when a job is marked complete manually, its stats are still collected. We’ve logged this as a feature request, and we’re currently targeting it for the 4.1 release.
Cheers,
great, more robust stats will help us a great deal. Our “small” group of 12 guys operates in a company of 80,000 and we have to do a lot of metric tracking to get new hardware.
in the mean time i am doing the following work around.
I mark the priority on the suspended jobs to 0
I then delete the output path so I dont overwrite “good” frames
I then requeue 1 task. the entire job then seems to go into the stats.