I have a peculiar behaviour of my ECS 2.0 when using qbittorent (QT app). It happens only for some torrent,s not for all. Qbittorrent tnds to stop its download with "stalled' result and then the whole ecs internet connection starts to break down. IM clients often lose messages, disconnect, cannot connect etc. Firtefox can load only begining of pages or cannot load at all and cannot be properly closed - have to be killed. All other internet apps have similar problems. Only reboot helps, although I think that in the past a very long wait sometimes helped.
The only clue that I have is that opened OpenSSH session during this network breakdown ended with message:
Write failed: No buffer space available
Anyone can helpme solving this issue?
Hi lewhoo
I've seen that behaviour with earlier builds of qt and qbittorrent but not with the current releases. What versions are you using?
Regards
Pete
true, it's not the latest: 2.6.7
As I was afraid, latest qbit torrent with latest qt - similar problems. It is not necessary for qbittorrent to go to "stalled" to break other apps internet connection...
Hi lewhoo
I've not seen this problem at all with qt473/qbittorrent285 - I often have qbittorrent running all the time the system is booted and no problems; currently running while I type this in Seamonkey.
Have you totally cleared out (deleted) *all* files from the earlier build of qt? - If the earlier build was installed using Warpin then it is best uninstalled using Warpin.
Maybe there is a setting that needs tweaking in the qbittorrent Preferences.
Maybe the patched tcpip32.dll is needed for qbittorrent, see http://www.os2world.com/content/view/21034/2/ - I'm using this and guess it could be the difference.
Sorry to leave you with more questions than answers... Hopefully 1 or more of the above will help :-)
Pete
I have also had problems with earlier releases of Qbittorrent, they frooze my system after some time running. However, the latest release has worked flawless. I then tried both with and without the patched tcpip32.dll which made no difference.
I see this occasionally (with the latest qbt and QT4.7). It seems to happen with torrents that have a very large number of fast peers. Personally I think the OS/2 IP stack just gets overwhelmed sometimes.
It may help to reduce the maximum number of simultaneous connections or uploads.
It also might help, when you run into a torrent that triggers this, to pause all other torrents while the problem one is running.
Hello, lewhoo et al.
It probably wouldn't hurt to limit the number of uploads/downloads, and the settings at which they can happen.
Until I laid down some restrictions, I experienced a bit of bandwidth hogging myself.
All this is under your Options>Connection and Options>Speed.
I limit my upload number to 4 and upload speed to 10. I do not restrict my downloads. (I also restrict the number of peers.) Those numbers work the best using my connection, your connection needs may vary somewhat.
May that act as a pointer towards your goal.
Thanks all. So far it seems to work, although I need longer time for testing. I deleted the qt config files in %HOME/local/share (or something like that). However, not all settings were deleted - it still knew where to download files (and there are people who claim that allowing apps to put their files where they want to does not lead to mess with which users have to deal...).
About the bandwidth - I definitely reduced the upload before, don't remember how about upload number. Well, I'll wait and see if with default settings the hogging will happen again.
Unfortunately the problem reappeared. I wonder - maybe it is connected with me chaning priorities of different files within the torrent, however it shouldn't. None of the options seems obviously connected to this behaviour.
The only thing I can think of is to uninstall QT and reinstall it....
Ugh... I have to dig. There's a post someplace on the forums about this, but certain TCP/IP stack settings will cause this issue. persist, timeout and keepalive from memory. Eventually the stack is handling a bit more than it can with those settings. The defaults, under one particular installation scheme (CP1 perhaps? or pre-CP1?) dont work very well for a server environment with lotsa connections - I think the same might apply to a bittorrent type app.
A *tiny* bit of info on it is at the link below. If I find the stuff I wrote up a while ago, I'll post it - otherwise, I'll try to check my settings later and post them.
http://www.os2world.com/os2files/os2/tuning/network.html
(keepalive value changed from 7800 to 60 - helps with all sorts of weird issues where the routing table starts getting massive).
inetcfg -s keepalive 60
I've found the info about timout earlier and it helped a lot I think. I hope I do not say it to early ;) Now there is only one app that has apparent problems and that's an old, ported IM communicator. It disconnects or receives mesages much later, or loses packets...
Anyway, it seems that also the number of sockets is flooded: netstat -s gives maaaany entries after Qbittorrent is running.
Here's a copy of my inetcfg.ini file.
Maybe it can be used as a guide for you.
#Inetcfg: CURRENT DEFAULT MINIMUM MAXIMUM
#A combination of two old ones. Modified 110206
keepalive 60 7800 0 7800 KeepAlive (sec)
tcpswinsize 32768 32768 1024 246723 TCP SendWindow Size
tcprwinsize 32768 32768 1024 246723 TCP RecvWindow Size
udpswinsize 9216 9216 1024 246723 UDP SendWindow Size
udprwinsize 41600 41600 1024 246723 UDP RecvWindow Size
lingertime 120 120 0 65535 Linger Time (sec)
probecnt 8 8 1 8 Number Of KeepAlive Probes
# IPForwarding is handle by IPGATE in Setup.cmd
# ipforward 1 0 0 1 IP Forwarding flag ON/OFF
mtudiscover 1 1 0 1 TCP Path MTU Discovery ON/OFF
arpkillc 1200 1200 180 15300 ARP Complete TimeOut (sec)
arpkilli 180 180 60 1200 ARP InComplete TimeOut (sec)
lipcstswinsize 10240 10240 1024 246723 LIPC Stream SendWindow Size
lipcstrwinsize 10240 10240 1024 246723 LIPC Stream RecvWindow Size
lipcdgswinsize 2048 2048 1024 246723 LIPC Dgram SendWindow Size
lipcdgrwinsize 4096 4096 1024 246723 LIPC Dgram RecvWindow Size
synattack 1 0 0 1 SYN Attack Flag ON/OFF
syncookie 1 0 0 1 SYN Cookie Flag ON/OFF
firewall 0 0 0 1 FIREWALL ON/OFF
multidefrt 0 1 0 1 Multiple Default Routes ON/OFF
winscale 1 1 0 1 TCP Window Scale ON/OFF
timestmp 1 0 0 1 TCP TimeStamps ON/OFF
realslow 20 20 1 100 TCP TW Q Slow Timeout ticks
perfhttp 1 1 0 1 Fast Path HTTP ON/OFF
tcpttl 64 64 1 255 TCP TTL (hops)
udpttl 64 64 1 255 UDP TTL (hops)
icmpttl 64 64 1 255 ICMP TTL (hops)
fragttl 60 60 1 255 Fragment TTL (sec)
reusetw 1 1 0 1 Reuse Time Wait ON/OFF
msl 30 30 1 16384 TCP MaxSegLifeTime (sec)
cc 0 0 0 1 TTCP function ON/OFF
sack 0 0 0 1 Selective ACK ON/OFF
# Selective ACK is said to be always left off; The reason why, I have lost.
I have this loaded upon startup with:
if exist %etc%\inetcfg.ini CALL inetcfg.exe -set all
using: b4tcp.cmd to load it.
Comments and/or suggests, welcome.