11 replies to this topic

[wantowan2]

I've been having this recurring issue after one of Apex's famous 'crashes'. This is over multi platform, multi version. After, there are always ports left open. Usually one of the UDP ports. Still happening with the latest Beta vers.

Not at all sure how to approach this. Setting in Apex? Router?

Help please

[Mek]

If ports are still open then Apex must be still running. I am pretty sure that the OS frees any handles used by a program after it is closed (normally or unexpectedly). Look in Task Manager when this happens, if you can see ApexDC(-x64).exe there.

[wantowan2]

Mek, on 11 August 2011 - 06:22 PM, said:

If ports are still open then Apex must be still running. I am pretty sure that the OS frees any handles used by a program after it is closed (normally or unexpectedly). Look in Task Manager when this happens, if you can see ApexDC(-x64).exe there.

Ya, that was my first thought as well. Unfortunately Apex is dead and gone 5-10 minutes sometimes and I still get an error message the next time I start Apex up stating that a UDP port is open and will effect performance. I test by trying to search and it's right, no results, the ports open... I think it''s the DHT/UDP port.
Even my regular hubs think that I am still logged on and tell me to pick a different Nick....lol
I've tried all kinds. Setting up port ranges instead so I can switch them while in the client to avoid having to reboot my router every time I launch Apex. (Did I mention it's Apex v.1.5.0.1000Beta X64)

[Mek]

when a client closes unexpectedly, most hubs still keep a "shadow" of it until certain timeout - that would explain the issue with nick already taken. However, about your port issue, I have no idea...  well... aren't you using some security software which would monitor traffic and think Apex is still running and hold the port open for it? I really don't know, just guessing here...

[wantowan2]

Mek, on 12 August 2011 - 06:33 PM, said:

when a client closes unexpectedly, most hubs still keep a "shadow" of it until certain timeout - that would explain the issue with nick already taken. However, about your port issue, I have no idea...  well... aren't you using some security software which would monitor traffic and think Apex is still running and hold the port open for it? I really don't know, just guessing here...

Yes, possible. Been having multiple issues. DNS probs w my R2 machine, I think my router is about fried and when I start Apex, half the time I get an error stating that it can't reach "The Server" (?) and shuts down Apex when I click ok on it?

Also, I have a ton of "time-outs" & "all download slots taken" in download queue. Haven't seen too many of those before....hmm

On top of that, I'm pretty new to this protocol...lol

One more.. I don't think that the GeoIp is resolving properly. Up until a day or so I could see all flags and IP's in DL-Q.. Now, only about 25% and I only seem to be able to up/download from those users..... Seems fishy?

Edited by wantowan2, 15 August 2011 - 03:22 AM.

[Mek]

You really have a lot of strange problems. About the GeoIP database, use the menu item Help -> GeoIP database update and extract the downloaded CSV file to ApexDC++ root directory (not under Settings or anywhere else) and restart Apex. Anyway, this does not have to do anything with your inability to download from users without flag... and that is weird.

[Crise]

Regarding the original problem, ports can be left open by upnp in the event of a program crash... which may result in further attempts at forwarding through upnp to fail. That said it should not cause the error message about port being in use to pop up.

This is something that should be resolved by the program, however, for now the best way to go about it would be to access the upnp rules by different method and clean the problematic rules. Alternatively you could look into manual port forwarding.

[wantowan2]

 Crise, on 17 August 2011 - 09:27 AM, said:

Regarding the original problem, ports can be left open by upnp in the event of a program crash... which may result in further attempts at forwarding through upnp to fail. That said it should not cause the error message about port being in use to pop up.

This is something that should be resolved by the program, however, for now the best way to go about it would be to access the upnp rules by different method and clean the problematic rules. Alternatively you could look into manual port forwarding.

Haha, that's the thing. All 4 ports forwarded. I even tried setting router up w ranges to make it easier. Still open sometimes. The default lease time for UDP ports (for me) is 5min. I tried 2.5min->75seconds just to see if that would help...lol Bahh

The problem with resolving user IP's is still a there. Likely/maybe it has nothing to do w the CSV file. Maybe it just looks that way? The CSV can (prob.?) only work with an IP present, correct? Makes sense that I can't DL from a user if for some reason I can't get there IP to show. Therefor, no flags. Correct?

As for the DNS issues, I've tried the Dyn-DNS trick instead of entering an actual 'outside' ip addy..... it works ok I guess. I get a new router/modem on Saturday so we'll see. meh...!

>>>>whisper<<<<  uTorrent & jDownloader still work fine. So my machine(s) must be ok.  But w DC++ you get the community aspect that's so cool. ;-)

Thanks for the help so far and if you think of anything else....

[Mek]

wantowan2, on 18 August 2011 - 03:48 AM, said:

The CSV can (prob.?) only work with an IP present, correct? Makes sense that I can't DL from a user if for some reason I can't get there IP to show. Therefor, no flags. Correct?

The CSV file and flags are only a display issue. You should still be able to access those users and still see their IP address but without a flag. If you can't see their IP address and it shows Connecting... all the time then you've got a connectivity issue.
Hope the new router sets everything right...

[wantowan2]

The included pic is of my current port config. I assume giving a range is ok. But are 4 ports necessary? Better/worse? Or is this why the 'DHT/UDP' port stays open?

Attached Files

•  Apex Ports1.jpg   102.68K   7 downloads

[Mek]

No, port range is not neccessary. Also, the port numbers may be the same as long as they differ in being TCP or UDP. From the picture you've posted, I understand you've mapped a whole outside port range to a single port on your host computer, and that's not needed

[wantowan2]

 Mek, on 23 August 2011 - 06:05 PM, said:

No, port range is not neccessary. Also, the port numbers may be the same as long as they differ in being TCP or UDP. From the picture you've posted, I understand you've mapped a whole outside port range to a single port on your host computer, and that's not needed

Ok, thats good. I put the range for my ISP (traffic shaping?) and other port assaults. Changing it now and then keeps PeerBlock quiet...


Ya, this is all a big mystery for me... My network was stable and Apex was running great for the first 2 days, then these connection probs.

Do you think or has there been much testing of Apex on Server 2008 R2? Is it considered simply win7 ult.?