Thread: [Linecontrol-development] Yet unknown behaviour...
Brought to you by:
sfuchs
Menu
▾
▴
linecontrol-development
|
From: Joao C. <jp...@rn...> - 2001-10-07 22:05:09
|
Hello. I've once said that I was unable to use one of the newer linesrv versions, because the line would connect and then disconnect after a couple of seconds... I have this behaviour in the last (2.1.4) version. My client connects , the line is up for some seconds, and then disconnects. I've made my isdn-up and isdn-down line write something in a log when they are called... And see what strange behaviour I get : --------------------------- dialed ippp0 at Sun Oct 7 21:00:29 WEST 2001 dialed ippp0 at Sun Oct 7 21:00:31 WEST 2001 Issued hangup at Sun Oct 7 21:00:40 WEST 2001 ( I added the ISP name to get some more info, but are the same logs ) dialed ippp0 (Iol) at Sun Oct 7 21:09:40 WEST 2001 dialed ippp0 (Iol) at Sun Oct 7 21:09:43 WEST 2001 Issued hangup at Sun Oct 7 21:09:49 WEST 2001 ------------------------------- Why the strange thing this issues 2 isdn-up calls? And why does the hangup is called next? I think this somekind of timeout related, but timeout of what? I have my config timeouts for 15 secs, so these are not it... I've reversed my network back to linesrv 2.1.0 , that is still stable for me... Can this behaviour be introduced by some missing option in my config file ? (I'm using the same in 2.1.4 as in 2.1.0) |
|
From: Joao C. <jp...@rn...> - 2001-10-07 22:10:15
|
Hello. I said some days ago that my WLC would fail as soon as I started it , with some error of acess violation or something.. It's happening the same with the new linesrv version, so it must not be a problem with linesrv but with the changes from the former versions of WLC. One curious thing : This fails in my Win2k box (as a super user) , and as I have some more win boxes, I decided to try out and i found that It runs perfect in Win95 boxes! So I tryed again in my win2k box , this time as Administrator (MS version of root?) and... it WORKED! I dunno if this brings some light to the problem but ...here goes... |
|
From: Joao C. <jp...@rn...> - 2001-10-07 23:46:49
|
Dump of my /var/log/daemon.log : Oct 7 20:37:54 lfsPentiumRouter linesrv[1996]: LineControl ML Server 2.1.4 started. Oct 7 20:38:05 lfsPentiumRouter linesrv[1996]: 192.168.1.8:1102 registered Oct 7 20:42:05 lfsPentiumRouter linesrv[1996]: cleanup(): cought signal 11 (...) Oct 7 20:42:15 lfsPentiumRouter linesrv[1996]: 1 lines are still active. Close them manually! Oct 7 20:42:15 lfsPentiumRouter linesrv[1996]: SEGFAULT, please report to sf...@gm... Oct 7 20:42:15 lfsPentiumRouter linesrv[1996]: Value of segf_indicator: 16 Oct 7 20:43:36 lfsPentiumRouter linesrv[2104]: LineControl ML Server 2.1.4 started. Oct 7 20:43:43 lfsPentiumRouter linesrv[2104]: 192.168.1.8:1104 registered Oct 7 20:43:49 lfsPentiumRouter linesrv[2104]: 192.168.1.8:1104 online Oct 7 20:44:19 lfsPentiumRouter linesrv[2104]: 192.168.1.8:1104 online (...) Oct 7 20:44:58 lfsPentiumRouter linesrv[2104]: cleanup(): cought signal 11 Oct 7 20:45:08 lfsPentiumRouter linesrv[2104]: 1 lines are still active. Close them manually! Oct 7 20:45:08 lfsPentiumRouter linesrv[2104]: SEGFAULT, please report to sf...@gm... Oct 7 20:45:08 lfsPentiumRouter linesrv[2104]: Value of segf_indicator: 16 Those 2 lines above where 192.168.1.8 gets online twice are related to that bug report I issued earlier. As we can see, there is no offline() called by the client in the middle. Each of those calls would call isdn-up twice and isdn-off once I don't see much info here but... About the segfault, this seems the problem introduceb by my change to get multiple ISP, as that "safety code" I cuted would mask whatever is the problem here by forcing the lines to close... I'm back to plain vannilla linesrv 2.1.0 for now, as it seems the more stable about this segfault Somehow I get the feeling that if the newer version have some work around this problem, it has been away from the real cause of it and it will get away eventually only in the new 3.x version... So maybe lets not bother too much about this! Greetings to you all and next weekend maybe I'll send you some more goodies :) «»«»«»«» Joao Clemente «» jp...@rn... «»«»«»«» Well... it's not really a leak, is it? The memory will be returned when the user reboots ... «»«»«»«» ( Maybe a known OS philosophy? ) «»«»«»«» |
|
From: S. F. <sf...@gm...> - 2001-10-08 08:16:07
|
>Oct 7 20:42:15 lfsPentiumRouter linesrv[1996]: Value of segf_indicator: 16 Good... that's what I need. Unfortunately there was bug until now (fixed in cvs as of Oct 8th 09:47 MET [+0200]). Could you take the actual cvs version and reproduce the segfault? The value 16 here is probable (90% or more) wrong... (jp: you don't have cvs working?... wait, you'll get a copy...) >Those 2 lines above where 192.168.1.8 gets online twice are related to that >bug report I issued earlier. >As we can see, there is no offline() called by the client in the middle. >Each of those calls would call isdn-up twice and isdn-off once >dialed ippp0 at Sun Oct 7 21:00:29 WEST 2001 >dialed ippp0 at Sun Oct 7 21:00:31 WEST 2001 >Issued hangup at Sun Oct 7 21:00:40 WEST 2001 These lines, are they coming from isdn4linux? If yes: to check how many times your script_up got called you have to add a line in the script_up: echo "script_up called: `date`" >> /var/log/updownstuff.log Anything else is crap for me and doesn't help. (sorry). >I don't see much info here but... >About the segfault, this seems the problem introduceb by my change to get >multiple ISP, as that >"safety code" I cuted would mask whatever is the problem here by forcing the >lines to close... I'm not going to bother about your changes because I didn't see your code. So if you report bugs, please use either cvs oder some version I released (at least: send me the code or a diff...) When you're talking about code: please notice the file and line you're talking about... I just can't catch up. >Somehow I get the feeling that if the newer version have some work around >this problem, it has been away from the real cause of it and it will get >away eventually only in the new 3.x version... So maybe lets not bother too >much about this! About 3: It will have new bugs and can't be compared to 2.x or earlier. It's written in C++ and has a completely different concept in it's code. Have a look at it and you'll see that... ;) Start it with "linesrv -d" if you wan't to see a list of all keywords, their defaults and format strings. Config parsing works. Now I'll have a look at the lcp3 implementation... >Can this behaviour be introduced by some missing option in my config file ? >(I'm using the same in 2.1.4 as in 2.1.0) Send me your config file... maybe you already sent it some weeks / months ago... but hey, I don't collect config files... ;) I answer the mail and then the whole thing gets deleted. The segfault bug: you're not the only one... but I currently don't have any installed isdncard or even analog modem. So the only way is to simulate the whole thing... that's not quite easy. But I would be happy to get the following thing: you're offline, isdn line is closed # cat /dev/isdninfo > somefile.iinfo & [1] <pid> # <dial line, online, hangup, closed > # kill <pid> # gz somefile.iinfo then mail me somefile.iinfo.gz This would allow me to write a isdninfo emulator (using a pipe) to debug isdn code of linesrv 3 (and 2.1.x too...) And if someone of you is using a strange isdn config (multiple cards, tons of channels, or whathever) please send me the same... ;) Greetings & thanks for your reports. Stefan Fuchs |
|
From: Joao C. <jp...@rn...> - 2001-10-08 00:15:18
|
I'm sorry , I think it is best to ignore the e-mail I send about getting isdn-up script runned twice. Dunno exactly what happened, but I changed to linesrv 2.1.0 and it did the same again (isdn-up twice and then isdn-off)! I removed the garbage from my scripts directory (the files isdn-on~ , isdn-off~ , and so on) , modified my /etc/ip-up, and a few more things that I cannot in any way relate with this problem, and... well, it seems to have disapeared, all linesrv versions seem to work fine now (part from the segfaults) Sorry to have botter you before, and let me post a fresh segfault I got just when my line came up as I was trying 2.1.4 : Oct 8 01:01:52 lfsPentiumRouter linesrv[853]: 192.168.1.9:2197 registered Oct 8 01:02:10 lfsPentiumRouter linesrv[853]: cleanup(): cought signal 11 Oct 8 01:02:20 lfsPentiumRouter linesrv[853]: 1 lines are still active. Close them manually! Oct 8 01:02:20 lfsPentiumRouter linesrv[853]: SEGFAULT, please report to sf...@gm... Oct 8 01:02:20 lfsPentiumRouter linesrv[853]: Value of segf_indicator: 16 This segfaulted exactly when it would write the "online" line (the line was open and kept open) Regards Joao Clemente |
Thanks for helping keep SourceForge clean.
X