Slow download after 98%

[Embryo 0]

I have a problem that is quite frustrating. I'm downloading with 4 to 6 MB\s until 98% when the download speed drops to 10 kb\s or so. This is especially frustrating when I'm downloading large files because the Estimated Completion Time drops from 1 minute to several hours. Please help me.

[Seeker 0]

There is nothing you can do (almost). When you get to 99% or similar, you only need that last allocated chunk to complete it. If a user with slow upload is assigned to upload you that last chunk you are stuck with it. Get it now?

What i do sometimes (if im arsed to do it) is, i "remove all sources" from a file and then "search for alternates" its all in the right-click menu when you click on a file. But usually i just wait for it to complete, i leave DC running all the time anyway so it doesent make any difference.

[Embryo 0]

I usually remove from the queue users that have a bad connection, but the problem is that, at that point, I can't even download from the fast users, it's like I'm stuck with the slowest ones.

[Seeker 0]

That is most likely purely coincidental. Dont worry about it, it will finish eventually.

[ifmn 0]

Close the connection to the slow user, apex will try and connect to a new one, maybe you'll download faster from him, if not - try another user...

[dagoburd 0]

Froum queue remove the slow user, and then add new user for who you know that has good speed to you :(

[Embryo 0]

Well, that seems to be the real problem, after 98% I can only download from the slow users and if I try to connect to the fast ones, I get the "No slots available" or "Disconnected" messages, and that is no coincidence because it always happens the same. I forgot to say that at that point Aphex removes all the sources I'm not downloading from, the ones with "No needed part", and I can't download from them again, I tried to search for them with "Search for alternates" but that doesn't seem to work.

[Seeker 0]

Its not Apex fault, if all the fastest users get their slots takken by someone else other then you. Tough luck dude! Also, ofcourse you wont find them again, couse their not assigned to upload you that last chunk you need.

1. Remove all sources

2. Search for alternates

(also hubsoft may allow you 1 search every 20seconds or even every 3 hours, all depends on which hub your in.)

Purpose of filesharing is to make you stay connected longer so you can actually upload some stuff. dont just leech, stay connected 'always' you will eventually get all the files, and upload some too.

[Embryo 0]

I'm no noob, I know all this things, but it's not quite pleasant when I'm downloading with 10 kb\s and uploading with 15 MB\s. Guess I'm really really unlucky, right? Well, thanks anyway for trying to help me, I'm still hoping this is a bug that will be fixed with a future patch.

[Zlobomir 3]

Well, try it with SDC++ (not very clever to send you to another client, but we do best for our users and for Apex in long terms ). If it persists:

1. It is not a bug.

2. BM has to fix it.

I wonder however, if it was a bug in Apex, why it happens only to you...? Can it be connected with type of files (less sources), less hubs?

[3dfx 2]

its true....no matter what the file format and/or the size,when it reaches 98% the download slows down....i have kept a close observation,and it just slows the download speed form the same sources.

i have experienced this in other DC clients thou,so i dont think its a ApexDC problem.

[Seeker 0]

Ok, we will look into this, on internal testing.

edit: did a bit of testing, failed to reproduce.

case 1: 10MB file, 4 users, 1 active, download of 70kB/ps, total of 5 chunks i believe, speed varied from beggining to end but was never lower then 40kB, sucessfull download.

case 2: 5MB file, 3 users, 3 active, download of 280kB/ps, 2 chunks, speed was constant (couse of only 2 chunks and small size file), sucesfull ligtning fast download

case 3: 360MB file, 48 users, 8 active, download of 520kB/ps, allot of chunks, constant speed from each user (more or less), segmented obviously works fine for those interested, didnt find any anomalies from 90% to 100% of completion.

[Noctis 0]

I had this problem with CZDC++ or StrongDC++ RC10 a long time ago. I was downloading a 2 disc movie (which quality was poor ), and the last 1 or 2 MiBs of the file took more than 2-3 days do download, however it only occured with one user, the other never had a slot open... ;)

Edited April 15, 2007 by Noctis

[Embryo 0]

I had this problem with CZDC++ or StrongDC++ RC10 a long time ago. I was downloading a 2 disc movie (which quality was poor ), and the last 1 or 2 MiBs of the file took more than 2-3 days do download, however it only occured with one user, the other never had a slot open...

I was starting to think that I was the only one with this problem. It can't be a network or hub error, because I also used other clients and never had this problem, and if testers say that they didn't find anything wrong, I wonder what could be the cause of the problem, or, most important, the solution.

[Zlobomir 3]

My guess is... TBI code.

[Seeker 0]

My guess is... TBI code.

Superseeding was removed from 0.5.alpha TBI's code isnt the issue.

[Zlobomir 3]

Hm, afaicr it is in 0.4.0, but not in 0.5.x, but I might be wrong...

[Lee 84]

Hm, afaicr it is in 0.4.0, but not in 0.5.x, but I might be wrong...

Correct.

[Embryo 0]

What is TBI code, could that be it ?

[ifmn 0]

No. It's superseeding feature...

[TheBlazingIcicle 0]

What, what? Accusations

I wonder how many of the people that you suddenly couldn't download from were partial sources? It is expected that in the randomness of many people downloading, there will end up some chunks that are very rare (bittorret has this too). Suddenly there are no partial sources anymore and you have to wait for a real source.

This was much worse in PWDC when the last chunk was always rare.

If I am right, look at the speed you get at the end, this is the speed you would get all the time with strongdc or any other client.

[Zlobomir 3]

Sorry TBI, I just assumed that your code is the only thing which is currently tested only by users... And this is so because it seemed to cause runtime errors on my machine, and afaicr Crise removed it for safety. Is there any dll dependencies, any known issues? My PC can't be so unique.

[Seeker 0]

What are you on about. He is talking about beta2, and superseeding was removed from 0.5.0, the code isnt an issue in this case. As for the code itself, it sort of worked didnt it? Maybe TBI can elaborate in detail why it was removed. One reason i know of is that he didnt have time to complete it.

[Zlobomir 3]

If by "he" you mean Embryo, I can't see any info that he is using the latest 1.0.0 B1 or B2, and I tactically didn't wanted to speak about the betas and encourage their spreading at all, still. :(

[Seeker 0]

If by "he" you mean Embryo, I can't see any info that he is using the latest 1.0.0 B1 or B2, and I tactically didn't wanted to speak about the betas and encourage their spreading at all, still.

rofl, betas are released on frontpage. How much more can you encourage them.. theyre ment to spread :P