Having search problem / connection problem?

Use this forum to flesh out your feature request before you enter it in <a href="http://dcpp.net/bugzilla/">Bugzilla</a>.

Moderator: Moderators

Locked
[Telia]Per(Swe)
Posts: 2
Joined: 2006-02-12 13:56

Having search problem / connection problem?

Post by [Telia]Per(Swe) » 2006-02-12 14:30

After browsing through this messageboard I realized that most DC++ users who are behind a router doesn't seem to understand a problem with all kind of routers. A problem that you will experience in the way that no new downloads starts from the download queue (if DC++ is already running), or that there's a timeout on all attempts to connect to any hub.

The reason for this problem has to do with how the routers are working. If there's too many connection attempts at the same time on a specific TCP/UDP port from the outside world, then the router regard this as a security risk and block that port. On older routers this simply caused the router to hangup and freeze on all ports, but on the routers that you can buy today it only block that specific port. It doesn't matter what kind of operating system you're using, nor which kind of Windows version or if you're in active or passive mode in DC++. Your router will simply block that port if there's too many connection attempts at the same time on the same port.

And, the only way to solve this problem is to disconnect the power supply to the router and re-connect it again, causing the router to free its memory from the information to block that port. You should always wait at least 5 seconds and don't just immediately turn it off and on. And you should also close DC++ and restart it.

I've used many different routers during the years on many different computers with differnt OS and Windows versions. And believe me, it doesn't matter which brand of router you're using. The problem is the same for them all. Although different routers seem to be more or less sensitive; it differs how many connection attempts from the outside world they regard as a security risk. The only way to make sure that this problem occur as seldom as possible is to make you less attractive to connect to, sharing as little as possible and connect to as few hubs as possible, and with as few upload slots as possible. However, this contradicts the basic idea of using P2P programs like DC++.

Xan1977
Forum Moderator
Posts: 627
Joined: 2003-06-05 20:15

Post by Xan1977 » 2006-02-12 14:47

You know that connecting to hubs and uploading runs on random outgoing ports, right? I don't disagree with you, but it's probably a result of the NAT table in the router overflowing. Most routers just die when that happens.

[Telia]Per(Swe)
Posts: 2
Joined: 2006-02-12 13:56

One single INCOMING TCP/UDP port, many random outgoing ditto

Post by [Telia]Per(Swe) » 2006-02-12 15:36

If you have a router and configured it to re-direct (port forwarding) INCOMING traffic on a specific port to that computer that you run DC++ on, and have configured DC++ to ACTIVE mode, then you have also configured DC++ to listen to that specific port. So all INCOMING traffic to DC++ goes through one specific port (TCP/UDP). It's true that OUTGOING traffic from DC++ runs on random ports, but that is not a problem. The problem is that DC++ only listens to one single port in active mode.

For you developers of coming versions of DC++, I suggest that you allow DC++ to be able to listen to a range of TCP/UDP ports in Active mode instead of just single port. Because most routers can be configured accordingly (to re-direct a range of ports to a specific computer within the LAN).

GargoyleMT
DC++ Contributor
Posts: 3212
Joined: 2003-01-07 21:46
Location: .pa.us

Re: One single INCOMING TCP/UDP port, many random outgoing d

Post by GargoyleMT » 2006-02-13 11:42

[Telia]Per(Swe) wrote:For you developers of coming versions of DC++, I suggest that you allow DC++ to be able to listen to a range of TCP/UDP ports in Active mode instead of just single port.
I'm not sure this is a good idea. And it's certainly a change from the current behavior, and it would break INFs in ADC. Most existing software listens on a single port, I'm not sure why routers have the option to forward ranges, since there is so little software (that I'm personally familiar with) that supports it.

PseudonympH
Forum Moderator
Posts: 366
Joined: 2004-03-06 02:46

Post by PseudonympH » 2006-02-13 14:55

BitTorrent used to, back when a new process was created for each torrent. Now, though...

bastya_elvtars
Posts: 164
Joined: 2005-01-06 08:39
Location: HU
Contact:

Post by bastya_elvtars » 2006-02-13 17:34

I just ask one thing: What would the benefits be?
Hey you, / Don't help them to bury the light... / Don't give in / Without a fight. (Pink Floyd)

Carraya
Posts: 112
Joined: 2004-09-21 11:43

Post by Carraya » 2006-02-14 03:09

Unless you have a really really high speed and a bad router I can't see any advantages at all...
<random funny comment>

GargoyleMT
DC++ Contributor
Posts: 3212
Joined: 2003-01-07 21:46
Location: .pa.us

Post by GargoyleMT » 2006-02-14 11:55

The way he phrased it, the benefit is from easier port forwarding by users. I just don't think that's a compelling reason, when everything else is based around single ports, not ranges. (Near as I can tell, the single port is probably justified by the exact same argument.)

Locked