I understand your core issue which is that incompletes aren't assembling. I'm unclear why that's happening to you.
For assemble incompletes to kick in:
1 - Has to have PAR files (files that exceed retries with no pars are failed to the failed list).
2 - Won't try until retries have expired (I use 2 retries).
3 - > 90% of the file blocks have to be present (file blocks, not PAR blocks).
The key is that retries gate Assemble incompletes. If all these things are true, it'll assemble incompletes and attempt a repair. So, knowing this, the question is why does it decide not to assemble? If you look at these incompletes that are hanging out, why do you think they're not assembling? More than 10% missing? Retries not expiring?
I'm not averse to removing #3. Download what I have, run the retries and if I have pars assemble and attempt repair. If you only use one server then you're sort of at their mercy anyway so retries are unlikely to fix anything.
When incompletes are assembled, they're padded.
Another issue is the passwords, as is, the lack of assigning passwords to a nzb set slows down things a lot, with ie a password file of 15+kb it is apparent that newsbin no longer directly tries the assigned password but rather all passwords in the list.
It should be possible to change this back to first use an assigned password then the list if needed.
We need to be able to verify assigned passwords, too. Having a column to display it directly is also a nice idea, but i guess overkill for the added comfort. If it isn't it would be really nice to see.
Do you actually have 15K of valid passwords in the list? Assigning passwords does nothing these days. The password is assigned using the first chunk of the first par file. There's no cracking. If you don't have a valid password in your list, it's going to be failed.
We need to be able to verify assigned passwords, too.
Any files with invalid passwords get kicked into the failed list as spam automatically. Unless you force the download that is.
Also an issue, because incomplete files are not written to disk as they used to, par repair takes longer as it prefers to fill a missing file completely rather than the chunks, making repair impossible unless a manual "assemble incomplete" is done.
Chunks on disk have nothing to do with par repair. Same for chunks in memory. Only completed files get scanned for par blocks. Where the chunks are is transparent to that process. Maybe you're talking about re-constituting files using just the PAR files.
SSD's aren't as delicate as you might think:
http://techreport.com/review/27062/the- ... fter-1-5pbI'm not averse to making changes for you. This just doesn't happen to me, probably because I use multiple servers.