Klever

Evgeny Novikov wrote:

Have we any obstacles for that?

I don't remember. I've created the branch "download_cancelled" where downloading cancelled jobs is allowed so Anton can experiment with his cancelled jobs.
A job has a lot of data, so if it is corrupted then for example Recalculation of cache can fail (it is happening when jobs are uploaded). Cancelling the job will corrupt its data in common case.

Evgeny Novikov wrote:

I think that this will be the main trouble. IMHO even for uploaded canceled and corrupted jobs caches should be recalculated in the way that is possible, i.e. all failures should be ignored. I think that this is already done for decided jobs since corresponding information (Safes, Unsafes, Unknowns, resources) is shown gracefully even for canceled and corrupted jobs. Let's wait for Anton feedback.

Cache is calculated during decision. But it can't be realculated manually (on manager's tools page) for cancelled and corrupted jobs.

Vladimir Gratinskiy wrote:

Evgeny Novikov wrote:

I think that this will be the main trouble. IMHO even for uploaded canceled and corrupted jobs caches should be recalculated in the way that is possible, i.e. all failures should be ignored. I think that this is already done for decided jobs since corresponding information (Safes, Unsafes, Unknowns, resources) is shown gracefully even for canceled and corrupted jobs. Let's wait for Anton feedback.

Cache is calculated during decision. But it can't be realculated manually (on manager's tools page) for cancelled and corrupted jobs.

As in #7525 this limitation looks really strange. We do need to refactor caches recalculation.