hiya all..I'm new here, I need some help :(

I'm using 4 different internet line with a load balancer. Jdownloader has been working great with 6 chunk and 4 downloads at a time. But after those chunk became 1/6, the download are getting slow due to my fastest speed internet finished the chunk 1st..its getting from 1MB/s to 16KB/s per file, so I need to stop and restart my downloads to make my fastest internet line downloading that chunk.

The problem is, I need to do it every 10 minute for 450MB file..anyone can suggest for a script or crontab (I'm using Ubuntu) or something to do this automatically?

Thanks in advance


edit: this is what I meant by that..

jD doesn't support dynamic chunking. It's better to increase the number of simultaneous downloads and set "Max. Con." to 1.

If you have multiple IP addresses you can try to use them simultaneously by running multiple instances of jD.

hi remi,

ya I tried that..but I got 512Kbps, 1Mbps, 2Mbps & 4Mbps line..I read somewhere here saying we need to set max con to 1 and # download to 20, but when my 512Kbps line start downloading 2 or 3 files, it gonna be forever to finish downloading except if I Stop download and Start download it again (so my 4Mbps will download it).

haa..I actually just solved the problem..took a few trials though

I make this crontab

*/10 * * * * /usr/bin/jdownloader --stop-download && sleep 1 && /usr/bin/jdownloader --start-download


btw, for those who using multiple IP addresses, I got 3 account on 3 different file hosting..Fileserve and Megaupload will give you problem with Host plugin error or something like that..but so far, I tried Filesonic and it never give a problem. Didnt try any other hosting though..

If you use one chunk per download and the speed drops for that download then this is caused by the host's servers. There's nothing jD can do about it.

If you're downloading large files try to find files that have been split into several parts or download them using a higher speed line.

Hi again remy,

sorry..you're right that maybe JDownloader didnt support dynamic chunk, but I think my 4 port load balancer will take care of that..

this is screenshot when I change # download to 1.

and this is when my crontab to stop and start download working

before crontab stop the download :

after crontab start the download again :

notice that the same file from 13KB/s went to 35KB/s after my faster internet line took care of the chunk.

So, if I set my max connection to 1 and max download to 20, my 512kb/s line which only give me around 50KB/s download speed might downloading 3 file which mean 50/3 = around 16.6KB/s per file.

Downloading 400mb file would cost me forever, as my other 3 faster line gonna finished its part according to their speed and stay idle after finished the download.

anyhow, my problem already solve :D

What I conclude after looking at those partial screen shots is the following :-

1) Running 1 download is faster than running 4 downloads simultaneously.

2) Before crontab stops the download, the accumulated download speed of your 4 downloads is faster than after stopping and restarting them.

I don't understand what you're doing. Stopping and starting a download will, statistically speaking, always be detrimental to the overall download performance of jD, because there is some period of time that nothing can be downloaded.

Originally Posted by remi

What I conclude after looking at those partial screen shots is the following :-

1) Running 1 download is faster than running 4 downloads simultaneously.

2) Before crontab stops the download, the accumulated download speed of your 4 downloads is faster than after stopping and restarting them.

I don't understand what you're doing. Stopping and starting a download will, statistically speaking, always be detrimental to the overall download performance of jD, because there is some period of time that nothing can be downloaded.

1. Thats when its 6/6 chunk, when 1/6 the speeds drop drastically
2. Thats because I screenshot it before all the downloads connect. That screenshot just to show 1/6 chunk speed before and after restart.

Based on your answer, I realized that you dont have any experience with load balancing multiple internet lines. My load balancer will start using my fastest internet line till the bandwith is full (Need to set total bandwith for each line), then it will start using the 2nd,3rd and so on.

So, after my fastest internet line finished all his job downloading the chunk..it stay idle, waiting new for a new connection..in this case, until all the chunks (using the slow lines) have finished download and JD start downloading the new files.

When my crontab stop all the download, my load balancer will detect that no bandwith has been use..and when my crontab command the JD to start download again, my load balancer will start using my fastest line to download until the bandwith maxed out..which means that my fastest line will start finishing the chunk leftover by my slow lines.

I hope you can understand that.

I understand the advantages of load balancing but your examples don't provide the evidence/proof that your load balancing system works.

What I was saying in my first reply was that you don't need it for jD. You can just use several instances of jD, one for each line, and let them use your bandwidth optimally.

hmmm..alright, I'll make a video later and post it on youtube to proof it.

My setup is like this :-


4 Different ISP
|
TP-Link TL-R488T (4 Wan 1 Lan Load Balancer)
| 192.168.1.x
Asus RT-N16
| 192.168.20.x
|-----|-----|-----|
PC PC PC PC

1. The load balancer will do the load balancing.
2. Asus RT-N16 will control QoS and provide the Gigabits LAN and 300MB/s Wifi connection.

So, the end user IP is fix and we can't choose which line to which JD even if I open 5 instances of JD.

You don't need to make that video for me because you'll not convince me that load balancing as you do it will optimise your 4 lines.

If you've 4 Internet connections from 4 different ISPs you'll also have 4 different IP addresses. jD will do the optimisation.

jD will do the optimization - you mean use 4 instances of jD or set max connection to 1 and number of downloads to 20?

guess, you'll never understand it.

geez, how hard to understand this?

make it like this,

1. You got 1 file with 16GB size, and split it into 16 smaller files with HJ split or something into 1GB each files.

2. You got 3 friend in different house with different internet lines with different speed. Including you, they'll be 4 house with 4 internet lines. You're the richer one with 4MB/s broadband, Your other 3 friends each got 2MB/s, 1MB/s and one of the poor fella only got 512KB/s line.

3. With jD setting of 1 max connection and 4 downloads, each one of you will be downloading 4 files each..means each of you will downloads 1GB.

4. After finished downloading all the files, everyone will give their portion to you to extract it into 1 big 16GB file again. But, you're the one with 4MB/s broadband of course will finished the download first, due to your 400+Kb/s speed. You even might finish download it within 1 hour.

But how long does your friend with 512KB/s line will finish downloading that 1GB file? It might took him 16 hours.

You will be idling, doing nothing, waiting for your friends to finish downloading right?

Unless you tell them to STOP download and give you the portion so you can START download it yourself coz you got the faster line, then you will just wasting your time there.


I hope you understand my logic,

thanks.


If you still not convince/understand this then I guess, its not my loss

Originally Posted by remi

If you have multiple IP addresses you can try to use them simultaneously by running multiple instances of jD.

This is the optimisation I was talking about. Since you're talking about different computers you can easily run one instance of jD per computer, each one using one of your Internet connections.

By trying to aggregate your Internet connections and trying to manage them all at once you lose flexibility and performance. Compare it to the efficiencies of free markets and the bureaucracy (and corruption), waste and inefficiencies of socialism/collectivism/communism.

[Jiaz]

next major version supports multiple proxies/ips at the same time. so i would wait for that version

current jdownloader does not have dynamic chunking = if the last chunk is slow then only stop/start might help to get higher speeds.
dynamic chunking is planed for our next downloadcore