Page 2 of 2

Re: Simultaneous Downloads

Posted: 03 Oct 2020, 19:54
by kingmotley
Hugbug,
There are quite a few different scenarios in which downloading from multiple nzbs would be greatly beneficial. Such as when you have multiple sources:
Source A: 5000 day retention (spotty completion rate), limited 10 mb/s, unlimited account
Source B: 1000 day retention, limited to 1000 mb/s, unlimited account
Source C: 5000 day retention (100% completion rate), limited to 1000 mb/s, block acount

Typically I would set source A and B as priority 0, and source C as priority 1.
Then if you have a queue as follows:
Item 1, age of 1500 days.
Item 2, age of 9 days.

The problem right now is that Nzbget will use source A to try and download item1 first, and is capped at the provider's limit of 10mb/s. Source B won't participate because the age of the file is outside of the provider's retention. Sabnzb would/does allow Source B to then begin download Item 2 while Source A and C continue to work on Item 1.

Also, less important is at the end of a particular file. Waiting for the last few articles to complete for Item 1 before beginning to work on Item2 will leave many connections idling even in a simple 1 source, 2 item scenario.

Re: Simultaneous Downloads

Posted: 03 Oct 2020, 20:13
by kingmotley
Another scenario is when you have sources that differing amounts of connections and/or the response rate of those connections vary. I often (almost always), have sources working on different items in my queue with sabnzb because the quicker of my sources has already downloaded everything it can from the items near the top of my queue while the rest of the sources are still working on them. And more often than not, if one source is having an issue with articles missing on an item, then other sources may as well, and article not available/found responses take very little bandwidth, but is time consuming.

In fact, in this scenario, it would be best if a single source could be working on multiple nzbs at the same time, so that you aren't alternating from connection starvation (high incomplete articles) to bandwidth starvation (high complete articles), but that can be resolved by having multiple sources most of the time. This is definitely a harder thing to balance correctly, but always working on multiple files at once would reduce the probability.