Emergency pause fails to protect
Posted: Thu Jun 22, 2017 5:08 pm
When the free space on the hard drive(s) where NewsBin data and downloads are located (DataPath, DownloadPath) runs out, NewsBin pauses, supposedly until the user can free more space - so that no downloaded files are lost or corrupted.
Unfortunately, it is not what happens.
I have DataPath and DownloadPath pointing to different hard drives.
If either of them runs out of free space, things go very wrong even though NewsBin "pauses" (still working in the background!!!) - many bad or zero length files are generated as a result.
When the DataPath drive is full, even when NewsBin is "paused", corrupt and mostly 0 length chunks are still generated, even though NewsBin is not downloading. I assume, these new chunks are created as a result of a background assembling/decoding. But they are 0 length because there is no room for them.
So when the user makes more free space on the drive, those 0 length chunks then are used to create files on the DownloadPath hard drive, resulting in corrupt or 0 length files.
I recently had such an episode with 6.80 beta 4 when because of a problem with password protected posts, there was a bottleneck and the DataPath drive was filled.
As the result many rar parts on the DownloadPath drive were damaged or zero length (correctly indicated in all uppercase letters) in many different posts - hundreds of bad files were generated.
When the free space runs out on the DownloadPath drive, things are not much better.
These situations perhaps are easier to avoid, but they happen from time to time. It happened again to me a couple of days ago: I found NewsBin paused, but when I made more free space, NewsBin generated (on the DownloadPath drive) many damaged or 0-length rar parts in two posts - even though they were downloaded correctly to the DataPath drive, then they were damaged while moving to the DownloadPath drive. So I had to download those damaged rar parts again unnecessarily.
BTW, Retry or Resume Download do not work in such situations. The correct approach is:
1. to close NewsBin.
2. to delete the damaged (uppercase) rar parts manually.
3. to start Newsbin again.
I think Newsbin should be fixed in such a way that no bad chunks or bad files in the download location were ever created - a better pausing is needed, without background processes. At least for such automatic emergency situations.
But perhaps better also when the user pauses NewsBin: if I pause it because I need all the computer resources for some critical task and NewsBin is still using cpu and data bandwidth for hard drive activities - it is wrong.
Also, once NewsBin pauses automatically for low free space emergency, it should not resume by itself after only one small file is moved by the user, because the same situation repeats again and again, The User has to pause NewsBin anyway.
After an emergency pause, NewsBin should resume only when the user decides it can.
BTW, is there a way to set some custom free space limit: to tell NewsBin to stop downloading when a specified free space threshold is reached?
I would like to leave some reserve to avoid such emergency situations in the first place.
Unfortunately, it is not what happens.
I have DataPath and DownloadPath pointing to different hard drives.
If either of them runs out of free space, things go very wrong even though NewsBin "pauses" (still working in the background!!!) - many bad or zero length files are generated as a result.
When the DataPath drive is full, even when NewsBin is "paused", corrupt and mostly 0 length chunks are still generated, even though NewsBin is not downloading. I assume, these new chunks are created as a result of a background assembling/decoding. But they are 0 length because there is no room for them.
So when the user makes more free space on the drive, those 0 length chunks then are used to create files on the DownloadPath hard drive, resulting in corrupt or 0 length files.
I recently had such an episode with 6.80 beta 4 when because of a problem with password protected posts, there was a bottleneck and the DataPath drive was filled.
As the result many rar parts on the DownloadPath drive were damaged or zero length (correctly indicated in all uppercase letters) in many different posts - hundreds of bad files were generated.
When the free space runs out on the DownloadPath drive, things are not much better.
These situations perhaps are easier to avoid, but they happen from time to time. It happened again to me a couple of days ago: I found NewsBin paused, but when I made more free space, NewsBin generated (on the DownloadPath drive) many damaged or 0-length rar parts in two posts - even though they were downloaded correctly to the DataPath drive, then they were damaged while moving to the DownloadPath drive. So I had to download those damaged rar parts again unnecessarily.
BTW, Retry or Resume Download do not work in such situations. The correct approach is:
1. to close NewsBin.
2. to delete the damaged (uppercase) rar parts manually.
3. to start Newsbin again.
I think Newsbin should be fixed in such a way that no bad chunks or bad files in the download location were ever created - a better pausing is needed, without background processes. At least for such automatic emergency situations.
But perhaps better also when the user pauses NewsBin: if I pause it because I need all the computer resources for some critical task and NewsBin is still using cpu and data bandwidth for hard drive activities - it is wrong.
Also, once NewsBin pauses automatically for low free space emergency, it should not resume by itself after only one small file is moved by the user, because the same situation repeats again and again, The User has to pause NewsBin anyway.
After an emergency pause, NewsBin should resume only when the user decides it can.
BTW, is there a way to set some custom free space limit: to tell NewsBin to stop downloading when a specified free space threshold is reached?
I would like to leave some reserve to avoid such emergency situations in the first place.